Electrophysiological Signatures in Global Cerebral Ischemia: Neuroprotection Via Chemogenetic Inhibition of CA1 Pyramidal Neurons in Rats

Abstract

Background

Although there has been limited research into the perturbation of electrophysiological activity in the brain after ischemia, the activity signatures during ischemia and reperfusion remain to be fully elucidated. We aim to comprehensively describe these electrophysiological signatures and interrogate their correlation with ischemic damage during global cerebral ischemia and reperfusion.

Methods and Results

We used the 4‐vessel occlusion method of inducing global cerebral ischemia in rats. We used in vivo electrophysiological techniques to simultaneously record single units, scalp electroencephalogram, and local field potentials in awake animals. Neuronal damage and astrocyte reactivation were examined by immunofluorescence, immunoblotting, and quantitative real‐time reverse‐transcription polymerase chain reaction under chemogenetic inhibition of glutamatergic neurons. Electroencephalogram/local field potentials power and phase‐amplitude coupling of the theta and low‐gamma bands were reduced during ischemia and the acute phase of reperfusion. The firing rate of single units was enhanced by ischemia–reperfusion, and the phase relationship between the local field potentials theta band and neuronal firing was altered. Precise inhibition of hippocampus CA1 pyramidal neuron hyperactivity by chemogenetics rescued the firing dysfunction, ischemic neuronal damage, and A1 astrocyte activation.

Conclusions

Our results provide a comprehensive description of the characteristics of electrophysiological activity that accompany ischemia–reperfusion and highlight the significance of this activity in ischemic damage.

Nonstandard Abbreviations and Acronyms

Gi

rAAV‐CaMKIIα‐hM4D(Gi)‐mCherry

NC

negative control

NS

narrow‐spiking

WS

wide‐spiking

Research Perspective.

What Is New?

• We provide a compelling description of electrophysiological changes from the onset of ischemia through to the following 3 days reperfusion in freely moving rats subject to global cerebral ischemia.

• We highlight the roles of increased neuronal excitability in ischemic damage by chemogenetic inactivation of CA1 pyramidal neurons.

What Question Should Be Addressed Next?

• Our results suggest these electrophysiological signatures may provide valuable information relevant to the development of new targets for the prognosis and treatment of brain ischemia.

• It is a helpful theoretical addition to the field but warrants further testing associated with behavioral functional outcomes.

Cardiac arrest is a serious cardiovascular disease, resulting in high mortality rate and severe morbidity worldwide. Postcardiac arrest brain injury, caused by global cerebral ischemia and subsequent brain reperfusion following resuscitation, is the leading cause of death and long‐term disability in patients with cardiac arrest. ¹ , ² Reperfusion after acute cerebral ischemia is accompanied by delayed neuronal death and associated brain damage especially in hippocampus. ³ , ⁴ , ⁵ Despite extensive knowledge of the cellular and molecular mechanisms responsible for neuronal dysfunction and death in reperfusion pathology, ⁶ , ⁷ it remains challenging to translate the identified molecules into clinical treatments. ⁸ Therefore, there is still an urgent medical need to develop novel neuroprotective strategies to protect against the risks of reperfusion after brain ischemia. Recent studies have shed light on the electrical activity in neurons and related oscillatory brain networks during brain ischemia. ⁹ , ¹⁰ , ¹¹ These electrophysiological signatures may provide information relevant for the development of new targets for the prognosis and treatment of brain ischemia.

Intricate cellular processes govern synaptic communication and the induction of postsynaptic electrical potentials, which integrate to produce changes in axonal discharge. Electrical currents propagate from neuron to neuron across the brain, forming networks of oscillatory activity. Changes in the oscillation patterns of electrical activity recorded from the scalp under pathological brain states can be linked to neuronal degeneration and pathology outcomes. ¹² , ¹³ Brain activity correlates with functional outcomes post stroke, ¹¹ , ¹⁴ , ¹⁵ and modulation of neuronal activity has been reported as a potential treatment for focal cerebral ischemia. ¹⁶ , ¹⁷ Electric activity is also demonstrated in a global cerebral ischemia model ⁹ and reported to have clinical significance in resuscitation after cardiac arrest. ¹⁰ Together, these suggest that disturbance in electrical activity may be a key determinant and important clinical marker for brain ischemia pathology. However, detailed information on the effects of cerebral ischemia on single‐unit neural activity and oscillatory networks, or on the potential for electrical therapeutic interventions for brain ischemia, is still lacking.

Direct recording of electrophysiological signals in vivo is the gold standard for measuring neuronal activity because of the excellent temporal resolution. ¹⁸ , ¹⁹ Penetrating depth electrodes can detect voltage signals at the single‐cell and single‐spike levels with high spatial resolution and can accurately target deep brain structures. ²⁰ , ²¹ , ²² This method has been used extensively in neuroscience research to record both extracellular action potentials (spikes) from individual neurons as well as the signals generated by neuronal populations (local field potentials [LFPs]) in both anesthetized and freely moving animals. ²³ , ²⁴ , ²⁵ However, in vivo electrophysiology has only rarely been used to investigate spatiotemporal changes in neuronal function in animal models of brain ischemia. Therefore, it may be anticipated that in vivo electrophysiology may produce valuable information about cerebral ischemia.

Here, we track and resolve the dynamics of the electrophysiological signatures in hippocampal CA1 region in a 4‐vessel occlusion model of cerebral ischemia in awake rats. We find that specific chemogenetic inhibition of CA1 pyramidal neuron hyperactivity at the acute stage rescued the abnormal neuronal firing, neuronal damage, and astrocyte reactivation induced by ischemia.

METHODS

The study was conducted according to Preferred Reporting Items in the Animal Research: Reporting of In Vivo Experiments guidelines ²⁶ for animal preclinical studies. Authors declare that all data and supporting materials support the findings of this study are available in the article. Further information is available from the corresponding author upon reasonable request.

Animals, Antibodies, and Viruses

All experiments were performed on male Sprague–Dawley rats. Rats weighing 170 to 210 g were used for studies with virus administration. Rats weighing 220 to 260 g were used for studies without virus administration. All animals were group housed on a 12 h/12 h light/dark cycle with access to food and water ad libitum. All experimental procedures were in accordance with the guidelines approved by the Institutional Animal Care and Use Committee of Xuzhou Medical University. Efforts were made to minimize potential pain and discomfort of the animals during experiments. After surgery, rats were returned to their home cage for recovery before further experiments. All animals were randomly allocated into different treatment groups and data were assessed under blinded conditions.

A total of 65 rats were involved in the electrophysiological study of the sham and ischemia groups. Among these rats, 12 rats were eliminated because rats died or did not meet the ischemia criteria. Recording of electrophysiological data for 10 sham rats and 14 ischemia rats was unsuccessful. As a result, 29 rats, including 12 sham rats and 17 ischemia rats, were successfully used for both the ischemia model and subsequent electrophysiology recordings. For the chemogenetic experiments, a total of 60 rats were involved in rAAV‐CaMKIIα‐hM4D(Gi)‐mCherry (Gi) and negative control (NC) administration. Among these rats, 8 Gi administration and 10 NC administration rats were excluded because rats died or did not meet ischemia criteria. Recording of electrophysiological data for 9 Gi administration and 11 NC administration rats was unsuccessful. As a result, 22 rats, including 12 with Gi administration and 10 with NC administration, were used for the entire experiment.

Rabbit polyclonal anti‐cleaved caspase‐3(ab2302) and rabbit monoclonalanti‐C3 (complement component 3) antibody (ab200999) were purchased from Abcam Biotechnology (Cambridge, MA). Rabbit monoclonal anti‐NeuN (12943S) was purchased from Cell Signaling Biotechnology (Danvers, MA, USA). Mouse monoclonal anti‐GFAP (glial fibrillary acidic protein; clone G‐A‐5, G3893) was purchased from Millipore Sigma (Darmstadt, Germany). Alexa Fluor 488 goat antirabbit/mouse IgG (H+L), and Alexa Fluor 594 goat anti‐rabbit/mouse IgG (H+L) and alkaline phosphatase‐conjugated secondary antibodies were purchased from Thermo Fisher Scientific (Waltham, MA, USA).

rAAV‐CaMKIIα‐hM4D(Gi)‐mCherry and rAAV‐CaMKIIα‐mCherry (serotypes 2/9) were purchased from BrainVTA (Wuhan, China).

Global Cerebral Ischemia

Fifteen minutes of transient cerebral ischemia was induced by occlusion of 4‐vessel occlusion, as described previously. ²⁷ , ²⁸ , ²⁹ , ³⁰ Briefly, rats were anesthetized with 5% isoflurane for induction and 1.5% to 2% isoflurane for maintenance, then the vertebral arteries were electrocauterized and the carotid arteries were isolated. Electrocauterized rats were allowed to recover for 24 hours with an overnight fast. On the following day, the isolated carotid arteries were occluded with aneurysm clips for 15 minutes to induce cerebral ischemia; blood flow was restored (reperfusion) by releasing the clips. A sham operation was performed with the same surgical procedures and electrocauterization but without occlusion of the carotid arteries. Rectal temperature was maintained at 37±0.5 °C during and after the ischemic insult. Rats that lost their righting reflex within 30 seconds and whose pupils were dilated with unresponsive to light during ischemia were deemed as successful. Rats that did not meet these criteria were excluded from our study.

Microelectrode Implantation

To record spikes and LFPs, microelectrodes with 16 single nichrome wires (single‐wire diameter: 25 μm; distance between wires: 200 μm; Kedou Brain‐Computer Technology, Suzhou, China) were implanted into the hippocampal CA1 subregion, as previously described. ³¹ For simultaneous electroencephalogram (EEG) recording, 2 of the single microelectrode wires were replaced with silver electrodes. Briefly, rats were anesthetized with 5% isoflurane for induction and positioned in a stereotaxic frame. Under 1.5% to 2% isoflurane for maintenance, the fur on the surface of the scalp from the midline of the orbits to the midpoint between the ears was removed and then a suitably sized hole was drilled through the skull above the right hippocampus for microelectrode implantation (−3.5 mm posterior to bregma [anteroposterior], −2 mm mediolateral). Then, the microelectrodes were positioned and lowered through the drilled hole to an average depth of 2.2 to 2.7 mm. Recordings were monitored during microelectrode implantation to ensure optimal placement within the hippocampal CA1 subregion. Two small screws were inserted into the skull as the reference electrode and connected to the ground. Dental acrylic was applied to seal the microelectrodes to the bone. After surgery, rats were housed individually for at least 1 week to recover before LFP/spike recordings were performed.

Spike/LFP/EEG Recordings

Spike and LFP recordings were performed in freely moving rats and were initiated when the rats had recovered from the implantation surgery (at least 1 week post surgery). To minimize potential spatial cues and novelty, animals were habituated to the recording room for 2 to 3 days before experimental recordings and electrophysiological recordings were conducted in the home cages. Signals were recorded at 1hour, 3hours, 6 hours, 1 day, and 2 days after 24our recovery from electrocauterization for both sham‐operated and ischemic rats. Spike/LFP and EEG signals were recorded during ischemia and the early stages of reperfusion. Spike, LFP, and EEG signals from sham and ischemic animals were recorded via the NeuroLego recording system.

For spike recordings, signals from the microelectrodes were sent to a headstage, amplified by a 16‐channel amplifier, and sampled at 30 kHz by an electrophysiological recording system (NeuroLego System, Jiangsu Brain Medical Technology, China).

For LFP and EEG recordings, signals were amplified, filtered at 0.1 to 300 Hz, and sampled at 1 kHz.

Animals with more than 3 channels for quantification throughout the recording period were included for further analysis.

Off‐Line Spike Sorting and Statistics of the Unit Data

Spikes were sorted from the raw data with Offline Sorter V4 software (Plexon). Principal component analysis was employed to distinguish different units. A unit was classified as a single unit if <0.75% of the interspike intervals were <1 milliseconds, as described in our previous studies. ³² , ³³ Units were classified as putative interneurons (narrow‐spiking [NS]) or putative excitatory neurons (wide‐spiking [WS]) on the basis of their spike waveform features (half‐valley width and peak‐to‐valley ratio). ³⁴ Average firing rates were calculated by dividing the total number of spikes by the duration of the recording session.

Analysis of LFP Signals

Similar to previous studies, the delta band was filtered at 0.5 to 4 Hz, the theta band at 4 to 12 Hz, the low‐gamma band at 35 to 55 Hz, and the high‐gamma band at 65 to 120 Hz. To analyze ischemia‐elicited changes in LFP power, we compared the LFP power between the sham and ischemia groups at the indicated time points (1 hour, 3 hours, 6 hours, 1 day, 2 days, 3 days).

Stereotactic Virus Injection

Rats were placed in a stereotactic apparatus (Kopf Instruments, USA) under isoflurane anesthesia and received injection of rAAV‐CaMKIIα‐hM4D(Gi)‐mCherry or rAAV‐CaMKIIα‐mCherry. After exposing the skull, a small craniotomy was made with a cranial drill. For better coverage of the entire dorsal CA1, rats were bilaterally microinjected at 2 sites per hemisphere. Site 1: anteroposterior −3.1 mm, mediolateral ±1.6 mm and 2.8 mm ventral (dorsalventral); Site 2: anteroposterior −3.8 mm, mediolateral ±2.6 mm, dorsalventral−2.8 mm. All stereotaxic coordinates are given relative to bregma. Viruses were injected at a rate of 0.1 μL/min, controlled by an injection syringe pump. After each injection, the needle was retained for 10 minutes before being slowly withdrawn from the brain. Three weeks later, the transient global cerebral ischemia/reperfusion model was induced by 4‐vessel occlusion occlusion.

Immunofluorescence

Rats were anesthetized with isoflurane and intracardially perfused with saline solution, followed by 4% paraformaldehyde in 0.1 M phosphate buffer. Brains were immediately collected for incubation in 4% paraformaldehyde overnight and were dehydrated successively in 15% and 30% sucrose at 4 °C. Brain blocks were frozen and sectioned with a freezing microtome (Leica CM 1950) into 20‐μm sections. Sections were saved at −20 °C in cryoprotectant solution. After rinsing for 30 minutes with 0.1 mol/L PBS, sections were blocked with 0.3% (vol/vol) Triton X‐100 (30 minutes) and 10% (wt/vol) normal goat serum (ZSJQB, Beijing, China) in 0.1 mol/L PBS for 1 hour at room temperature. Then sections were incubated at 4 °C overnight with primary antibodies in 0.15% Triton X‐100 and 10% normal goat serum in PBS. After washing with PBS for 10 minutes×3, samples were incubated with Alexa Fluor 594‐ or Alexa Fluor 488‐conjugated secondary antibodies for 1 to 2 hours at room temperature followed by 4, 6‐diamidino‐2‐phenylindole staining. Finally, sections were washed in PBS for 10 minutes×4 at room temperature and mounted with Vectashield antifade mounting medium (Vector Labs, VA, USA). Confocal images were captured by a Fluoview confocal microscope (Zeiss, LSM 710). All sections were imaged with the same acquisition parameters and subjected to intensity analysis in ImageJ software.

Immunoblotting

As described previously, CA1 hippocampal tissues were harvested at the indicated reperfusion time points (6 hours, 1 day, 3 days, 5 days) after global ischemia. ³⁰ CA1 samples were homogenized in ice‐cold lysis buffer and quantified via the Lowry method. Identical amounts of protein samples solubilized in 4×Laemmli sample buffer were separated by SDS‐PAGE and then electrotransferred onto a nitrocellulose membrane (pore size: 0.2 μm). After blocking in 3% BSA, the membrane was incubated with the indicated primary antibody overnight. Detection was carried out by appropriate horseradish peroxidase‐conjugated IgG and developed with a SuperSignal West Pico Chemiluminescent Substrate assay kit (#34087) (Thermo Fisher Scientific). The protein bands were scanned and analyzed in ImageJ Software.

Real‐Time Quantitative Polymerase Chain Reaction Analysis

Total RNA was extracted from hippocampal CA1 tissues with TRIzol reagent (Invitrogen) following the manufacturer's instructions. RNA was reverse transcribed into cDNA with the HiScript III 1st Strand cDNA Synthesis Kit (+gDNA wiper) (Cat# R312‐01, Vazyme, Nanjing, China). Real‐time polymerase chain reaction (PCR) was performed on a Step One Plus Real‐Time PCR system (Applied Biosystems) with ChamQ SYBR qPCR Master Mix (Low ROX Premixed) (Cat# Q331‐02, Vazyme). All the reactions were repeated at least 3 times. Relative mRNA expressions were quantified by the 2^−ΔΔCt method, and the data were normalized to the β‐actin gene as an internal control. The primer pairs are listed in the Table 1.

Table 1.

Primer Pairs

Genes	Forward	Reverse
Amigo2	GTTCGCCACAACAACATCAC	GTTTCTGCAAGTGGGAGAGC
Fbln5	TTCCAGATGCAAGCAACGA	AGGCAGTGTCAGAGGCCTTA
Gbp2	GCTCTTAAACTTCGGGAACAGG	GTTTGGGCCTCGGACCTTTA
Ggta1	CTCTCAGGATCTGGGAGTTGGA	AAGCAAACAGCAGAGCAACC
Serping1	GCTCAGAGGCTAACTGGCTT	AGAAGGCTCTATCCCCAGCTA
Srgn	AATGGGTCCGCTGTAAACCA	TCGGGAATCCTCTCATCAAAAC
H2‐T23	AGAAGGCATGAAAAGACAGTTGC	GCAATAGGAAACACCCAGCC
β‐Actin	ACACCCGCCACCAGTTCG	ACCCATACCCACCATCACAC

Statistical Analysis

Electrophysiological data were analyzed in MATLAB 2020b. The Gaussian distribution of the data was assessed by the Anderson–Darling test. If the data sets were normally distributed, the data were tested for significance with a 2‐sample t test. If the data sets were nonnormally distributed, the data were tested for significance with a Wilcoxon rank sum test. The Benjamini‐Hochberg method was applied to calculate the adjusted P value for multiple comparisons. ³⁵ Statistical analysis of protein intensity, quantitative real‐time reverse transcription PCR, and immunoflurorescence were conducted in Graphpad Prism 7.0 software. Two‐sample t tests were conducted when comparing between 2 groups. When multiple group comparisons were required, a 2‐way ANOVA was performed followed by Tukey's test to determine group differences.

RESULTS

Global Cerebral Ischemia and Reperfusion Decrease Scalp EEG Power in Awake Animals

Noninvasive EEG records the overall activity from a pool of neurons and is a widespread technique for studying brain activity under physiological and pathological conditions. ¹⁸ To explore how ischemia condition modify brain oscillations, we recorded the scalp EEG signal in awake experimental animals. We made continuous recordings of scalp EEG signals at 1‐hour, 3‐hour, 6‐hour, 1‐day, 2‐day, and 3‐day reperfusion after ischemia or at corresponding time points after 24‐hour recovery from electrocauterization in sham groups. Each recording session was at least 40 minutes.

Figure 1A and 1B show example raw EEG traces from 1 sham rat and 1 ischemic rat at the indicated time points. Figure 1C depicts the normalized power in ischemic and sham rats at all EEG frequency spectra. Different frequencies of oscillation may carry distinct information important for different neural processes. We focused our analysis on the delta (0.5–4 Hz), the theta (4–12 Hz), low gamma (36–65 Hz), and high‐gamma (66–95 Hz) oscillations because they are important for encoding spatial information and consolidating memory. ³⁶ , ³⁷ , ³⁸ We found that global ischemia and reperfusion caused a reduction in delta EEG activity during ischemia, 1‐hour and 6‐hour reperfusion (Figure 1D, Wilcoxon rank sum test: *P=0.015; *P=0.049; *P=0.012. False discovery rate (FDR) correction: P=0.060; P=0.10; P=0.060). There was a persistent reduction in theta EEG activity from the onset of ischemia through 1 day of reperfusion (Figure 1E, Wilcoxon rank sum test: **P=0.0020; **P=0.0058; *P=0.012; **P=0.0049; 2‐sample test: *P=0.037. FDR correction: *P=0.016; *P=0.016; *P=0.024; *P=0.016; P=0.059). For the low‐gamma band, a similar decrease persisted from the onset of ischemia through 6‐hour reperfusion (Figure 1F, Wilcoxon rank sum test: *P=0.015; *P=0.026; *P=0.015; **P=0.0039. FDR correction: *P=0.040; P=0.051; *P=0.040; *P=0.031). The high‐gamma EEG was significantly reduced in ischemic animals only during the ischemia stage and at 6‐hour reperfusion (Figure 1G, Wilcoxon rank sum test: **P=0.0016; **P=0.0062. FDR correction: *P=0.013; *P=0.025). These results verify that global cerebral ischemia perturbs the scalp EEG in conscious rats.

Figure 1. Global cerebral ischemia–reperfusion suppresses scalp EEG activity in freely moving rats.

A and B, Representative scalp EEG traces recorded in sham control (A) and ischemic (B) rats preoperatively (BL), intraoperatively (SH or IS) periods, and at different postoperative time points. From top to bottom: Raw scalp EEG (top), bandpass‐filtered traces (middle; delta [δ], theta [θ], low gamma [γ], and high gamma [γ]), and the corresponding spectrogram (bottom). C, Average scalp EEG power spectra of sham control (black) and ischemic (red) rats preoperatively (BL), intraoperatively (SH/IS), and at different postoperative time points. Data are shown as the mean±SEM. D, Comparison of average power in the delta band in the sham (black) and ischemic (red) groups preoperatively (BL), intraoperatively (SH/IS), and at different postoperative periods. (From left to right: Wilcoxon rank sum test: Z=0.11, P=0.92; Wilcoxon rank sum test: Z=2.43, *P=0.015; Wilcoxon rank sum test: Z=1.97, *P=0.049; Wilcoxon rank sum test: Z=1.52, P=0.13; Wilcoxon rank sum test: Z=2.51, *P=0.012; Wilcoxon rank sum test: Z=1.31, P=0.19; Wilcoxon rank sum test: Z=1.90, P=0.057; Wilcoxon rank sum test: Z=−1.86, P=0.063. FDR corrected P=0.92, 0.060, 0.10, 0.17, 0.060, 0.22, 0.10, 0.10). E, Same as (D), but for the theta band. (From left to right: 2‐sample t test: t₍₁₉₎=0.39, P=0.70; Wilcoxon rank sum test: Z=3.086, **P=0.0020; Wilcoxon rank sum test: Z=2.76, **P=0.0058; Wilcoxon rank sum test: Z=2.51, *P=0.012; Wilcoxon rank sum test: Z=2.81, **P=0.0049; 2‐sample t test: t₍₁₄₎=2.30, *P=0.037; Wilcoxon rank sum test: Z=1.81, P=0.070; 2‐sample t test: t₍₁₅₎=0.71, P=0.49. FDR corrected P=0.70, 0.016, 0.016, 0.024, 0.016, 0.059, 0.093, 0.56). F, Same as (D), but for the low‐gamma band. (From left to right: Wilcoxon rank sum test: Z=1.58, P=0.11; Wilcoxon rank sum test: Z=2.43, *P=0.015; Wilcoxon rank sum test: Z=2.23, *P=0.026; Wilcoxon rank sum test: Z=2.43, *P=0.015; Wilcoxon rank sum test: Z=2.89, **P=0.0039; 2‐sample t test: t₍₁₄₎=2.074, P=0.057; Wilcoxon rank sum test: Z=1.54, P=0.124; Wilcoxon rank sum test: Z=0.55, P=0.58. FDR corrected P=0.14, 0.040, 0.051, 0.040, 0.031, 0.091, 0.14, 0.58). G, Same as (D), but for the high‐gamma band. (From left to right: 2‐sample t test: t₍₁₉₎=1.15, P=0.26; Wilcoxon rank sum test: Z=3.15, **P=0.0016; Wilcoxon rank sum test: Z=1.84, P=0.066; Wilcoxon rank sum test: Z=1.82, P=0.068; Wilcoxon rank sum test: Z=2.74, **P=0.0062; 2‐sample t test: t₍₁₄₎=2.051, P=0.059; Wilcoxon rank sum test: Z=1.72, P=0.085; 2‐sample t test: t₍₁₅₎=1.75, P=0.10. FDR corrected P=0.26, 0.013, 0.11, 0.11, 0.025, 0.11, 0.11, 0.11). BL indicates baseline; EEG, electroencephalogram; FDR, false discovery rate; IS, ischemia; and SH, sham.

Global Cerebral Ischemia and Reperfusion Impair the Hippocampal Neuronal Network in Awake Animals

Whereas the scalp EEG signal represents activity over a large area of the brain, LFPs reflect the activity of more spatially restricted cell ensembles. ¹⁸ The hippocampus is one of the brain regions most vulnerable to global cerebral ischemia. To investigate specific changes in hippocampal neuronal network activity after ischemia, we recorded LFPs from hippocampal CA1. LFP signals were recorded in freely moving rats via 16‐channel microelectrodes stereotactically implanted in hippocampal CA1.

Figure 2A and 2B show examples of the raw and filtered (delta: 0.5–4 Hz; theta: 4–12 Hz; low gamma: 36–65 Hz; and high gamma: 66–95 Hz) hippocampal CA1 LFP traces from 1 sham and 1 ischemic rat at the reperfusion time points indicated. Figure 2C shows the overall changes in LFP rhythmicity in sham and ischemic rats: the amplitude of the hippocampal LFP was decreased dramatically in the ischemia group compared with the sham group, especially at the acute stage. We next conducted a power spectral analysis of the delta, theta, low gamma, and high gamma bands to compare the sham and ischemia groups at the indicated time points. The average power of the delta was significantly decreased during the ischemia and 1‐hour reperfusion, compared with the sham (Figure 2D) (Wilcoxon rank sum test: *P=0.023 and **P=0.0088. FDR correction: P=0.094; P=0.070). For the theta band, it was also significantly decreased during the ischemia (Wilcoxon rank sum test: *P=0.015. FDR correction: P=0.058) and 1‐hour reperfusion (Wilcoxon rank‐ sum test: **P=0.0016. FDR correction: *P=0.013) (Figure 2E). A reduction was detected in the low‐gamma and high‐gamma frequency bands at 1‐hour reperfusion (Wilcoxon rank sum test: **P=0.0022 in Figure 2F and Wilcoxon rank sum test: **P=0.0067 in Figure 2G. FDR correction: *P=0.017; P=0.054) and 3‐hour reperfusion (Wilcoxon rank sum test: *P=0.024 in Figure 2F and Wilcoxon rank sum test: *P=0.039 in Figure 2G. FDR correction: P=0.095; P=0.15) compared with the corresponding sham rats. These results indicate that reperfusion after cerebral ischemia decreases neural network activity in the hippocampus of awake rats, especially in the delta, theta, low‐gamma, and high‐gamma frequency bands.

Figure 2. Global cerebral ischemia–reperfusion suppresses dorsal hippocampal CA1 LFP activity in freely moving rats.

A and B, Representative LFP traces recorded in sham control (A) and ischemic (B) rats preoperatively (BL), intraoperatively (SH or IS), and at different postoperative time points. From top to bottom: Raw LFP (top), bandpass‐filtered traces (middle; δ, θ, low γ, and high γ), and the corresponding spectrogram (bottom). C, Average LFP power spectra of sham control (black) and ischemic (red) rats preoperatively (BL), intraoperatively (SH/IS), and at different postoperative time points. Data are shown as the mean±SEM. D, Comparison of average power in the delta band in the sham (black) and ischemic (red) groups preoperatively (BL), intraoperatively (SH/IS), and at different postoperative time points. (From left to right: Wilcoxon rank sum test: Z=1.22, P=0.22; Wilcoxon rank sum test: Z=2.27, *P=0.023; Wilcoxon rank sum test: Z=2.62, **P=0.0088; Wilcoxon rank sum test: Z=1.59, P=0.11; Wilcoxon rank sum test: Z=1.49, P=0.14; Wilcoxon rank sum test: Z=1.79, P=0.074; Wilcoxon rank sum test: Z=0.87, P=0.39; Two‐sample t test: t(12)=1.90, P=0.082. FDR corrected P=0.25, 0.094, 0.070, 0.18, 0.18, 0.16, 0.39, 0.16). E, Same as (D), but for the theta band. (From left to right: 2‐sample t test: t₍₁₅₎=1.88, P=0.079; Wilcoxon rank sum test: Z=2.44, *P=0.015; Wilcoxon rank sum test: Z=3.15, **P=0.0016; Wilcoxon rank sum test: Z=1.40, P=0.16; Wilcoxon rank sum test: Z=1.78, P=0.075; Wilcoxon rank sum test: Z=1.79, P=0.074; Wilcoxon rank sum test: Z=0.52, P=0.60; 2‐sample t test: t₍₁₂₎=1.20, P=0.25. FDR corrected P=0.13, 0.058, 0.013, 0.22, 0.13, 0.13, 0.60, 0.29). F, Same as (D), but for the low‐gamma band. (From left to right: 2‐sample t test: t₍₁₅₎=2.06, P=0.057; Wilcoxon rank sum test: Z=1.29, P=0.20; Wilcoxon rank sum test: Z=3.066, **P=0.0022; Wilcoxon rank sum test: Z=2.26, *P=0.024; Wilcoxon rank sum test: Z=1.40, P=0.16; 2‐sample t‐test: t₍₁₂₎=1.35, P=0.20; Wilcoxon rank sum test: Z=0.75, P=0.45; Wilcoxon rank sum test: Z=0.58, P=0.56. FDR corrected P=0.15, 0.27, 0.017, 0.095, 0.27, 0.27, 0.52, 0.56). G, Same as (D), but for the high‐gamma band. (From left to right: 2‐sample t test: t₍₁₅₎=1.63, P=0.12; Wilcoxon rank sum test: Z=1.47, P=0.14; Wilcoxon rank sum test: Z=2.71, **P=0.0067; Wilcoxon rank sum test: Z=2.069, *P=0.039; Wilcoxon rank sum test: Z=0.34, P=0.74; 2‐sample t test: t₍₁₂₎=1.036, P=0.32; Wilcoxon rank sum test: Z=0.64, P=0.52; Wilcoxon rank sum test: Z=0.32, P=0.75. FDR corrected P=0.29, 0.29, 0.054, 0.15, 0.75, 0.51, 0.70, 0.75). BL indicates baseline; FDR, false discovery rate; IS, ischemia; LFP, local field potentials; and SH, sham.

Cross‐frequency coupling (CFC) of LFP oscillations in different frequency bandwidths can further enhance the information‐carrying capacity of a neural network. ³⁹ Theta and gamma oscillations in the hippocampus are both linked to memory and phase‐amplitude coupling (PAC) of the theta and gamma signals has been reported in the hippocampus but has not been thoroughly characterized. ⁴⁰ , ⁴¹ , ⁴² To investigate whether global ischemia and reperfusion quantitatively change the CFC of hippocampal LFPs, we performed PAC analysis for theta phase and low‐gamma amplitude from the hippocampal CA1 LFP data. Figure 3 shows quantification of PAC intensity by the modulation index, a measure of how localized the amplitude of low gamma oscillations is to the phase of the theta oscillations. Figure 3A1 and 3A2 show examples of the extracellular LFP sampled and filtered between 1 and 120 Hz for a sham rat and an ischemic rat. Theta amplitude and instantaneous phase were extracted from the Hilbert transform of the filtered signal (Figure 3B and 3C). We used PAC analysis to evaluate the degree of coupling between the envelope for low‐gamma LFP amplitude and the phase of the theta LFP. The blue lines in Figure 3D show the envelope for the amplitude of the low‐gamma LFP, calculated with the Hilbert transform. Figure 3E shows the phase‐amplitude plot of the normalized distribution of low‐gamma amplitudes in different theta phase bins. The PAC strength was estimated as the modulation index. Statistical analysis of the mean modulation index for ischemia and sham groups revealed a significant decrease in modulation index at 1 hour reperfusion (Figure 3F, Wilcoxon rank sum test: **P=0.0067. FDR correction: P=0.054). Thus, these results suggest a tendency for theta phase‐gamma amplitude coupling to be decreased by global cerebral ischemia and reperfusion.

Figure 3. Global cerebral ischemia–reperfusion decreases cross‐frequency coupling in the dorsal hippocampal CA1 LFP.

A, Representative raw LFP traces recorded in sham control (A1) and ischemic (A2) rats at the 6‐hour reperfusion time point. B, Theta‐band LFP trace derived from the raw LFP trace in (A). C, Theta phase after Hilbert transform of the raw LFP trace in (A). D, Low‐gamma amplitude envelope after Hilbert transform of the raw LFP trace in (A). E, Left: theta phase distribution for low‐gamma amplitude from an individual recording. Right: polar histogram of the high‐gamma peak theta angles for all recordings. F, Comparison of mean MI for the sham (black) and ischemic (red) groups preoperatively (BL), intraoperatively (SH/IS), and at different postoperative time points. (From left to right: Wilcoxon rank sum test: Z=1.12, P=0.26; Wilcoxon rank sum test: Z=0.29, P=0.77; Wilcoxon rank sum test: Z=2.71, **P=0.0067; 2‐sample t test: t₍₁₅₎=1.90, P=0.077;Wilcoxon rank sum test: Z=−0.14, P=0.89; 2‐sample t test: t₍₁₃₎=1.25, P=0.23; Wilcoxon rank sum test: Z=−0.37, P=0.71; Wilcoxon rank sum test: Z=0.84, P=0.40. FDR corrected P=0.52, 0.88, 0.054, 0.31, 0.89, 0.52, 0.88, 0.64). BL indicates baseline; FDR, false discovery rate; IS, ischemia; LFP, local field potentials; MI, modulation index; and SH, sham.

Global Cerebral Ischemia and Reperfusion Cause Hippocampal Hyperactivity in Awake Animals

To investigate the neural basis of the observed network dysfunction at the individual‐neuron level, we performed single‐unit spike recording in the hippocampus during ischemia and reperfusion in awake rats (Figure 4A). Immunohistochemistry was performed on postmortem brain sections at the end of the study to confirm electrode placement (Figure 4D). Representative firing traces of hippocampal neurons in sham and ischemia groups were shown in Figure 4B and 4C, and single units could be sorted from the raw spikes (Figure 4E; see Methods for details of spike isolation). The mean firing rate was higher in the ischemia group than in the sham group during the acute reperfusion stage (1 hour, 3 hours, 6 hours, 1 day, 2 days) (Figure 4F, Wilcoxon rank sum test: ***P<0.001; **P=0.0040; **P=0.0078; P=0.15; ***P<0.001. FDR correction: ***P<0.001; *P=0.011; *P=0.016; ***P<0.001). Previous studies indicate that NS and WS cells likely correspond to putative interneurons and putative excitatory cells, respectively. ³⁴ For NS cells, firing rates were increased at 1 hour, 6 hour, and 2 days reperfusion versus sham rats (Figure 4G, Wilcoxon rank sum test: ***P<0.001; *P=0.013; ***P<0.001. FDR correction: ***P<0.001; *P=0.035; ***P<0.001). In WS cells, firing rates were increased at 1‐hour, 3‐hour, and 2‐day reperfusion versus sham rats (Figure 4H, Wilcoxon rank sum test: *P=0.014; **P=0.0013; **P=0.0021. FDR correction: *P=0.036; **P=0.0086; **P=0.0086). These results from single‐unit recordings indicate hyperactivity of both putative interneurons and putative excitatory neurons in hippocampus during reperfusion after global ischemia.

Figure 4. Global cerebral ischemia–reperfusion induces single‐unit hyperactivity in dorsal hippocampal CA1 neurons in awake rats.

A, Experimental timeline for electrophysiological recordings. B and C, Representative raw spike traces recorded from sham control (B) and ischemic (C) rats preoperatively (BL), intraoperatively (SH or IS), and at different postoperative time points. D, Left: schematic diagram of the microelectrode array. Right: schematic of a hippocampal slice showing placement of the microelectrodes. Scale bar: 400 μm. E, Example of spike sorting using principal components analysis to identify clusters in the extracellular voltage recorded by the microelectrodes. Here, the spike sorting resulted in identification of two separate units. F, Comparison of the MFR for all recorded neurons in the sham (black) and ischemic (red) groups preoperatively (BL), intraoperatively (SH/IS), and at different postoperative time points. (From left to right: Wilcoxon rank sum test: Z=1.0063, P=0.31; Wilcoxon rank sum test: Z=0.78, P=0.43; Wilcoxon rank sum test: Z=4.26, ***P<0.001; Wilcoxon rank sum test: Z=2.88, **P=0.0040; Wilcoxon rank sum test: Z=2.66, **P=0.0078; Wilcoxon rank sum test: Z=1.45, P=0.15; Wilcoxon rank sum test: Z=4.68, ***P<0.001; Wilcoxon rank sum test: Z=0.093, P=0.93. FDR corrected P=0.42, 0.50, 0.000081, 0.011, 0.016, 0.24, 0.000023, 0.93). G, Same as (F), but for narrow‐spiking units. (From left to right: Wilcoxon rank sum test: Z=−0.45, P=0.65; Wilcoxon rank sum test: Z=−0.39, P=0.69; Wilcoxon rank sum test: Z=3.61, ***P=0.00030; Wilcoxon rank sum test: Z=0.21, P=0.83; Wilcoxon rank sum test: Z=2.48, *P=0.013; Wilcoxon rank sum test: Z=0.81, P=0.42; Wilcoxon rank sum test: Z=3.97, ***P<0.001; Wilcoxon rank sum test: Z=−0.71, P=0.48. FDR corrected P=0.65, 0.65, 0.00067, 0.65, 0.035, 0.65, 0.00059, 0.65). H, Same as (F), but for wide‐spiking neurons. (From left to right: Wilcoxon rank sum test: Z=1.37, P=0.17; Wilcoxon rank sum test: Z=1.14, P=0.25; Wilcoxon rank sum test: Z=2.47, *P=0.014; Wilcoxon rank sum test: Z=3.21, **P=0.0013; Wilcoxon rank sum test: Z=1.75, P=0.079; Wilcoxon rank sum test: Z=1.27, P=0.20; Wilcoxon rank sum test: Z=3.069, **P=0.0021; Wilcoxon rank sum test: Z=0.60, P=0.55. FDR corrected P=0.27, 0.29, 0.036, 0.0086, 0.16, 0.27, 0.0086, 0.55). BL indicates baseline; FDR, false discovery rate; IS, ischemia; MFR, mean firing rate; and SH, sham.

To assess the temporal relationship between oscillatory rhythms and neuronal firing, we analyzed changes in spike‐field coherence during reperfusion (Figure 5). After calculating the theta phase with the Hilbert transform and plotting the single‐unit discharge in a raster diagram (Figure 5C and 5D), we calculated the mean firing rate per unit at a particular theta phase (Figure 5E). Testing for differences in the phase relationship between the neuronal firing and theta LFP rhythms revealed that overall spike‐field coherence for the theta LFP decreased during ischemia and 1‐hour reperfusion but increased at 3‐day reperfusion (Figure 5F, Wilcoxon rank sum test: ***P<0.001; *P=0.017; **P=0.0079. FDR correction: ***P<0.001; *P=0.045; *P=0.032). Specifically, coherence between NS activity and rhythmic theta oscillations significantly decreased during ischemia (Figure 5G, Wilcoxon rank sum test: ***P<0.001. FDR correction: ***P<0.001), and phase‐locking of WS activity and theta oscillations significantly decreased at 1‐hour but increased at 3‐day reperfusion (Figure 5H, Wilcoxon rank sum test: ***P=0.00052; ***P<0.001. FDR correction: **P=0.0021; ***P<0.001). These results indicate that global cerebral ischemia and reperfusion decrease the spike‐field coherence in the hippocampus at early stage but increase it at late stage.

Figure 5. Global cerebral ischemia–reperfusion affects phase coupling between spikes and LFP in dorsal hippocampal CA1.

A, Representative raw LFP traces recorded in sham control (A1) and ischemic (A2) rats after surgery. B, Theta‐band LFP trace derived from the raw LFP trace shown in (A). C, Theta phase after Hilbert transform of the raw LFP trace in (A). D, Raster plot showing the firing of one example single unit from each group. E, Left: theta phase distribution for the MFR of a single unit. Right: polar histogram of the firing‐rate peak theta angle for the population of recorded units. F, Comparison of theta phase precession in all recorded neurons in sham (black) and ischemic (red) groups preoperatively (BL), intraoperatively (SH/IS), and at different postoperative time points. (From left to right: Wilcoxon rank sum test: Z=0.61, P=0.54; Wilcoxon rank sum test: Z=4.05, ***P<0.001; Wilcoxon rank sum test: Z=2.39, *P=0.017; Wilcoxon rank sum test: Z=0.28, P=0.78; Wilcoxon rank sum test: Z=0.43, P=0.67; Wilcoxon rank sum test: Z=−1.18, P=0.24; Wilcoxon rank sum test: Z=−0.28, P=0.78; Wilcoxon rank sum test: Z=−2.66, **P=0.0079. FDR corrected P=0.78, 0.00041, 0.045, 0.78, 0.78, 0.48, 0.78, 0.032). G, Same as (F), but for narrow‐spiking neurons. (From left to right: Wilcoxon rank sum test: Z=0.90, P=0.37; Wilcoxon rank sum test: Z=4.28, ***P<0.001; Wilcoxon rank sum test: Z=0.49, P=0.63; Wilcoxon rank sum test: Z=0.64, P=0.52; Wilcoxon rank sum test: Z=0.17, P=0.86; Wilcoxon rank sum test: Z=−0.22, P=0.82; Wilcoxon rank sum test: Z=1.18, P=0.24; Wilcoxon rank sum test: Z=−0.14, P=0.89. FDR corrected P=0.89, 0.00015, 0.89, 0.89, 0.89, 0.89, 0.89, 0.89). H, Same as (F), but for wide‐spiking neurons. (From left to right: Wilcoxon rank sum test: Z=−0.68, P=0.50; Wilcoxon rank sum test: Z=0.95, P=0.34; Wilcoxon rank sum test: Z=3.47, ***P=0.00052; Wilcoxon rank sum test: Z=−0.46, P=0.65; Wilcoxon rank sum test: Z=0.63, P=0.53; Wilcoxon rank sum test: Z=−1.83, P=0.067; Wilcoxon rank sum test: Z=−1.48, P=0.14; Wilcoxon rank sum test: Z=−4.20, ***P<0.001. FDR corrected P=0.60, 0.55, 0.0021, 0.65, 0.60, 0.18, 0.28, 0.00022). BL indicates baseline; FDR, false discovery rate; IS, ischemia; LFP, local field potentials; MFR, mean firing rate; PLV, phase locking value; and SH, sham.

Chemogenetic Attenuation of Hippocampal Neuronal Activity Rescues the Abnormal Firing Rates Induced by Global Cerebral Ischemia and Reperfusion

Previous studies have used selective manipulation of neuronal activity at chronic ¹⁶ and acute ¹⁷ stages to improve recovery in focal ischemia models, such as middle cerebral artery occlusion. However, there are no reports of direct, specific manipulation of ischemia‐sensitive regions in global ischemia models. Our in vivo recording data described previously suggest that global cerebral ischemia and reperfusion cause neuronal hyperactivity and network dysfunction in the CA1 region of the hippocampus.

We used a chemogenetic approach to test whether direct in vivo manipulation of hippocampal neurons could alter the ischemia‐induced hyperactivity. A modified form of the human muscarinic M4 receptor under the CaMKIIα (calcium/calmodulin‐dependent protein kinase type II subunit alpha) promoter was virally delivered to the hippocampus, where it was selectively expressed in glutamatergic neurons. Activation of the human muscarinic M4 receptors was initiated 30 minutes after reperfusion by intraperitoneal clozapine‐n‐oxide injections (0.33 mg/100 g) to inhibit hippocampal neuronal activity (Figure 6A and 6B). In ischemic rats, this specific inactivation of glutamatergic neurons in the hippocampus effectively reduced the overall firing rate during reperfusion compared with the control virus (mCherry) with the same clozapine‐n‐oxide administration (Figure 6C) (Wilcoxon rank sum test: **P=0.0027; ***P=0.00089; *P=0.038. FDR vorrection: **P=0.0094; **P=0.0063; P=0.089). Further analysis of neuronal subtypes showed that this suppression of firing rate was specific only to WS cells (Figure 6E, Wilcoxon rank sum test: *P=0.045; *P=0.027; *P=0.023. FDR correction: P=0.11; P=0.094; P=0.094) but not NS cells (Figure 6D). These data indicate that inhibition of excitatory neurons shortly after reperfusion reliably ameliorates subsequent abnormal neuronal firing rates in vivo, providing a valuable basis for investigating the role of hippocampal activity in damage from global ischemia.

Figure 6. Chemogenetic inhibition of hippocampal neurons rescues abnormal firing rates induced by global cerebral ischemia–reperfusion.

A, Experimental time course of the viral injection, ischemia–reperfusion, and electrophysiological recordings. Created in BioRender. Liu, P. (2024) https://BioRender.com/m48p725. B, Hippocampal section showing placement of the microelectrodes and expression of hM4Di (Gi)‐mCherry. Scale bar: 400 μm. C, Comparison of MFR in all recorded neurons in rats injected with NC (red) or hM4Di (Gi) (blue) preoperatively (BL), intraoperatively (SH/IS), and at different postoperative time points. (From left to right: Wilcoxon rank sum test: Z=−0.87, P=0.38; Wilcoxon rank sum test: Z=0.73, P=0.47; Wilcoxon rank sum test: Z=1.73, P=0.083; Wilcoxon rank sum test: Z=1.30, P=0.19; Wilcoxon rank sum test: Z=3.00, **P=0.0027; Wilcoxon rank sum test: Z=3.32, ***P=0.00090; Wilcoxon rank sum test: Z=2.07, *P=0.038. FDR corrected P=0.45, 0.47, 0.14, 0.27, 0.0094, 0.0063, 0.089). D, As in (C), but for narrow‐spiking neurons. (From left to right: Wilcoxon rank sum test: Z=−0.58, P=0.56; Wilcoxon rank sum test: Z=−0.47, P=0.64; Wilcoxon rank sum test: Z=1.00, P=0.31; Wilcoxon rank sum test: Z=−0.24, P=0.81; Wilcoxon rank sum test: Z=0.90, P=0.37; Wilcoxon rank sum test: Z=1.80, P=0.072; Wilcoxon rank sum test: Z=−0.22, P=0.83. FDR corrected P=0.83, 0.83, 0.83, 0.83, 0.83, 0.51, 0.83). E, As in (C), but for wide‐spiking neurons. (From left to right: Wilcoxon rank sum test: Z=−1.37, P=0.17; Wilcoxon rank sum test: Z=1.67, P=0.094; Wilcoxon rank sum test: Z=1.45, P=0.15; Wilcoxon rank sum test: Z=1.14, P=0.25; Wilcoxon rank sum test: Z=2.00, *P=0.045; Wilcoxon rank sum test: Z=2.21, *P=0.027; Wilcoxon rank sum test: Z=2.27, *P=0.023. FDR corrected P=0.20, 0.16, 0.20, 0.25, 0.11, 0.094, 0.094). BL indicates baseline; CNO, clozapine‐n‐oxide; DAPI, 4, 6‐diamidino‐2‐phenylindole; FDR, false discovery rate; Gi, Gi, rAAV‐CaMKIIα‐hM4D(Gi)‐mCherry; IS, ischemia; LFP, local field potentials; MFR, mean firing rate; NC, negative control, rAAV‐CaMKIIα‐mCherry; and SH, sham.

Chemogenetic Attenuation of Hippocampal Neuronal Activity Alleviates the Neuronal Damage Induced by Global Cerebral Ischemia and Reperfusion

We next tested whether precise modulation of glutamatergic neurons during the acute phase of reperfusion after global cerebral ischemia could ameliorate the delayed and specific neuron cell death in hippocampal CA1. At the 5‐day reperfusion time point, the number of NeuN (neuronal nuclei)‐positive cells, indicating neuron survival, was markedly increased in rats expressing human muscarinic M4 receptors (plus clozapine‐n‐oxide treatment) compared with rats that received the control virus plus clozapine‐n‐oxide treatment (Figure 7A and 7B) (2‐sample t test: *P=0.030). To validate the neuronal loss, we evaluated the expression of the apoptosis‐related protein cleaved caspase 3 in the hippocampus. At the 1‐day reperfusion time point, the immunostaining signal of cleaved caspase 3 was lower in the human muscarinic M4 receptor‐expressing group than in the control group (Figure 7C and 7D) (2‐sample t test: **P=0.0037). Together, these data suggest that chemogenetic attenuation of hippocampal neuronal activity shortly after reperfusion reduces apoptosis and improves neuronal survival in the hippocampal CA1 subregion. This suggests that hyperactivity in excitatory neurons can be specifically targeted at the acute stage to arrest pathological neuronal damage during global ischemia and reperfusion.

Figure 7. Chemogenetic inhibition of hippocampal excitatory neurons rescues neuronal damage induced by global cerebral ischemia–reperfusion.

A, Representative images of NeuN immunohistochemical staining. Scale bar, 20 μm. B, Comparison of the number of NeuN‐positive cells in the NC and chemogenetic (Gi) groups. Two‐sample t test: t₍₈₎=2.64, n=5, *P=0.030. C, Representative photomicrographs of cleaved caspase‐3 (green) staining, indicating apoptosis. Scale bar, 50 μm. D, Comparison of the number of cleaved caspase‐3‐positive cells. Two‐sample t test: t₍₉₎=3.88, n=5 for NC and n=6 for Gi, **P=0.0037. GFAP indicates glial fibrillary acidic protein; Gi, rAAV‐CaMKIIα‐hM4D(Gi)‐mCherry; I/R, ischemia/reperfusion; NC, negative control, rAAV‐CaMKIIα‐mCherry; and NeuN, neuronal nuclei.

Chemogenetic Attenuation of Hippocampal Neuronal Activity Relieves Neurotoxic Reactive A1‐Like Astrocyte Activation in Ischemic Rats

As a dominant nonneuronal cell population, astrocytes are immediately activated after the onset of cerebral ischemia. ³⁰ , ⁴³ Studies indicate that astrocytes are functionally polarized into 2 classes: the A1‐like subtype that exerts neurotoxic effects and the A2‐like subtype that has neuroprotective effects. Each is respectively marked by a series of specific gene transcripts. ⁴⁴ Generation of neurotoxic A1‐like astrocytes could theoretically be initiated in focal ischemic rats, and reducing A1‐like astrocyte neurotoxicity may be a candidate target for therapeutic intervention. ⁴⁵ , ⁴⁶

We aimed to test whether global cerebral ischemia‐induced changes in astrocyte reactivation and phenotypic transition of A1‐like astrocytes are affected by chemogenetic attenuation of hippocampal neuronal activity. First, we confirmed an increase in reactive astrocytes during reperfusion after transient global cerebral ischemia by performing western blotting of CA1 tissue with anti‐GFAP antibodies. The GFAP band was significantly more intense at reperfusion for tissue from the ischemic group than the sham group (Figure 8A and 8B). As a common marker of A1‐like astrocytes, we used C3 to demonstrate whether neurotoxic reactive astrocytes were induced. In Figure 8A and 8B, the C3 immunoblotting signal was stronger in CA1 homogenates from rats subjected to ischemia/reperfusion than in sham rats. Next, we used quantitative real‐time reverse transcription‐PCR to examine mRNA levels of additional neurotoxic A1‐like astrocyte markers at the 6‐hour and 3‐day reperfusion time point. There was an obvious increase in mRNA levels of neurotoxic markers, including Amigo2, Fbln5, Gbp2, Gata1, H2‐T23, Seripng‐1, and Srgn, during reperfusion (Figure 8C and 8E). These results suggest that A1‐like astrocytes are induced during reperfusion after global ischemia. Notably, the increased GFAP immunoreactivity was significantly attenuated in the CaMKIIα‐hM4D (Gi) group (in which neuronal excitability was inhibited) but not in the control CaMKIIα‐mCherry group (Figure 8G through 8L). Furthermore, immunoblotting analysis of C3 and quantitative real‐time reverse transcription‐PCR analysis of other neurotoxic A1‐like astrocyte markers demonstrated that chemogenetic attenuation of neuronal activity restrained the increases in C3 protein and neurotoxic‐marker mRNA, whereas CaMKIIα‐mCherry did not have this effect (Figure 8D, 8F, and 8I through 8L). Together, our findings suggest that neurotoxic A1‐like astrocyte transition after global cerebral ischemia/reperfusion could be suppressed by chemogenetic attenuation of hippocampal neuronal activity.

Figure 8. Chemogenetic attenuation of hippocampal excitatory neurons decreases astrocyte reactivity and transition to A1‐like astrocytes during global ischemia–reperfusion.

A, Representative immunoblotting of anti‐GFAP and anti‐C3 in hippocampal CA1 from ischemia–reperfusion and sham‐treated rats. B, Quantification of the density of GFAP and C3 bands for (A). The relative levels were calculated by dividing the density of GFAP/C3 bands by that of the corresponding GAPDH band and then normalizing to the sham group. Data are expressed as the mean±SEM. n=7; *P<0.05, **P<0.01, ***P<0.001 vs the sham group. C and E, qRT‐PCR analysis of relative mRNA levels of 7 target genes for A1‐like astrocytes in hippocampal CA1 from reperfusion and sham rats. Data are the mean±SEM. n=3; *P<0.05, **P<0.01, ***P<0.001 vs the sham group. D and F, qRT‐PCR analysis of relative mRNA levels of 7 target genes for A1‐like astrocytes in hippocampal CA1 from reperfusion rats with or without chemogenetic modulation. Data are the mean±SEM. n=3; *P<0.05, **P<0.01, ***P<0.001 vs the sham group. G, Representative images of immunofluorescence staining for GFAP (green) from 1‐day reperfusion rats with or without chemogenetic modulation. H, Quantification of GFAP immunofluorescence. Data are expressed as the mean±SEM. Gi: n=4, NC: n=5; *P<0.05, **P<0.01, ***P<0.001 vs the sham group. I and K, Representative immunoblotting analysis of anti‐GFAP and anti‐C3 in hippocampal CA1 from sham rats and from reperfusion rats with or without chemogenetic modulation. J and L, Quantification of GFAP and C3 bands for (I) and (K), respectively. Data are expressed as the mean±SEM. n=3; *P<0.05, **P<0.01, ***P<0.001 vs the sham group. GFAP indicates glial fibrillary acidic protein; IB, immunoblotting; I/R, ischemia/reperfusion; Gi, rAAV‐CaMKIIα‐hM4D(Gi)‐mCherry; IB, immunoblotting; NC, negative control, rAAV‐CaMKIIα‐mCherry; and qRT‐PCR, quantitative real‐time reverse‐transcription polymerase chain reaction.

DISCUSSION

Although perturbations of electrical activity in the brain after ischemia have been reported previously, the detailed development of such electrophysiological changes in awake animals, and the functional consequences, have not been well characterized. Here, we used in vivo electrophysiology to record a spectrum of temporal neuronal activity patterns—EEG, LFPs, and single‐unit spiking—in hippocampal CA1 during ischemia onset and at multiple time points during reperfusion, and delved into the correlations between neuronal electrical activity and ischemic damage using chemogenetic techniques. We observed decreases in EEG/LFP power and PAC for theta and low‐gamma LFP during the acute phase of reperfusion after ischemia, as well as enhanced single‐unit firing. The phase relationship between the theta LFP band and neuronal firing was decreased at acute stage but increased at late stage. Importantly, specific inhibition of CA1 pyramidal neuron hyperactivity by chemogenetic techniques rescued the ischemia‐induced dysfunction in single‐unit firing rates and neuronal damage and suppressed A1‐like astrocyte activation. These findings provide direct evidence for dysfunction of hippocampal neurons during ischemia and reperfusion and identify a relationship between electrophysiological activity and ischemic damage. Our data broaden useful insights into the pathophysiological effects of brain ischemia. These results deepen our understanding of how brain ischemia modulates ongoing dynamics in the hippocampus at the level of single neurons and neural populations.

To understand the temporal development of ischemic pathophysiology, it is important to characterize neuronal activity in vivo and at a high temporal resolution during the acute phase. LFP signals are inherently ambiguous because they are comprised of contributions from multiple sources, making them harder to interpret, ²² but they can be stably recorded in a chronic setting. By contrast, spikes are extracellularly recorded somatic action potentials generated by individual neurons as the aggregate of synaptic membrane potentials and reflect information processing at the millisecond timescale. ²¹ To monitor the neural activity of large ensembles of neurons relatively completely, that activity needs to be measured simultaneously both as spikes and, over chronic time scales, as LFPs. Furthermore, high‐resolution recordings with depth microelectrodes targeting specific regions ²⁰ , ²¹ , ²² combined with advanced data processing methods can provide insights into the cooperative behavior of neurons and increase our understanding of dynamics in neuronal activity. ²⁵ Our comprehensive in vivo monitoring and analysis of LFPs and single‐unit spikes in hippocampal CA1 in awake rats provide additional information about electrophysiological signatures to complement previous findings. ⁹ In particular, our findings shed light on the association between neuronal activity in this specific region of the hippocampus and ischemia‐induced impairments. Our study suggests that manipulating neural electrical activity may be an alternative strategy for neuroprotective treatment in cerebral ischemia.

Because it can be recorded noninvasively, EEG is a commonly monitored electrophysiological signal and has been previously studied. Here, we found decreased EEG power, especially in the delta, theta, and low‐gamma bands, in awake rats during acute ischemia and subsequent early reperfusion (Figure 1). A previous study has reported that the amplitude of low‐gamma and high‐gamma LFP signals in hippocampal CA1 is dampened in a hippocampal ischemia model under anesthesia. ⁴⁷ Recently, significantly reduced power in the CA1 low‐gamma LFP band, but not the theta or high‐gamma band, was observed in awake mice with the 2‐vessel occlusion model of ischemia. ⁹ In the present study, LFP power in CA1 was reduced in the delta and theta through around 1‐hour reperfusion, and in the low‐gamma and high‐gamma bands through around 3‐hour reperfusion (Figure 2). This inconsistency between our results and those of the previous studies is likely attributable to the differences in ischemia models, animal state, and recording time points. In our study, we used the 4‐vessel occlusion method of inducing ischemia in rats, which leads to a more complete blockade of blood supply than other methods. We recorded the LFP signal from immediately after ischemia onset to 3 days reperfusion to obtain temporal data with which to dissect the gradual changes in electrical activity that occur during global cerebral ischemia and reperfusion. To make the data robust, we recorded LFPs for at least 40 minutes at each time point and included at least 8 rats in each group. It is worth noting that our study used conscious animals, avoiding the complications from anesthesia. General anesthesia with diverse anesthetics is known to affect EEG and LFP signals. ⁴⁸ , ⁴⁹ CFC between the theta and gamma LFP bands is the most comprehensive example of coupling of neural oscillations and has primarily been investigated in memory systems. ⁴¹ , ⁵⁰ , ⁵¹ , ⁵² Changes in PAC are the most common type of CFC change observed in neurological disorders. ⁵³ , ⁵⁴ , ⁵⁵ Interestingly, we found that coupling of hippocampal LFP theta phase and low‐gamma amplitude was decreased at 1‐hour and 3‐hour reperfusion in the ischemic group compared with the sham control (Figure 3). Poor hippocampal coupling between theta and gamma could predict poor memory performance. ⁵⁶ Then, we speculate that this altered CFC in the hippocampus may contribute to memory impairments associated with global cerebral ischemia.

Whereas oscillations in the EEG or LFP reflect network activity in a population of neurons, single‐unit spikes show the neural activity in individual cells. The spatiotemporally resolved activity of single neurons holds the key to understanding the inner workings of the brain. ¹⁸ Our results provide direct in vivo evidence of single‐unit hyperactivity in the hippocampus during reperfusion (Figure 4F). Both WS neurons and NS neurons exhibited increases in their mean firing rate during reperfusion (Figure 4G and 4H). Our results extend previous findings by examining perturbations in the activity of putative excitatory and inhibitory neurons in the hippocampus and by identifying changes in the coherence of individual spikes with theta‐band LFP during reperfusion. Changes in coherence were observed in both putative excitatory (WS) cells and putative interneurons (NS cells) (Figure 5), suggesting that both cell types have abnormal temporal modulation of spikes by theta oscillations after ischemia. Spike‐theta coherence plays a crucial role in memory formation, spatial navigation, information encoding, and sequential learning. ⁵⁷ , ⁵⁸ , ⁵⁹ The increase in spike‐theta coherence reflect enhanced synchronization. ⁶⁰ We speculate that this increased or decreased synchronization could be related to the reorganization of neuronal networks in the hippocampal region, serving as a response of the brain to ischemic injury.

However, the temporal dynamics of ischemia induced changes are not similar between firing and LFPs (Figures 2 and 4). It is reasonable to suppose that these measurements may reflect different aspects of the ischemic cascade while they both indicate neural network dysfunction. This might be due to the different impacts of ischemia on various types of neurons and neural networks, as well as the complex regulation of LFP oscillations. Additionally, the changes in single spike data following ischemia are highly dynamic and complex. In the early stages of ischemia, neurons rapidly undergo a series of acute pathological responses. At the 3‐day time point, the changes we observed may be related to complex processes such as cell death, tissue repair, and inflammatory responses. These changes may reflect the cumulative effects of the injury and the tissue's adaptation to it. For example, between 2 hours and 3 days, compensatory mechanisms may temporarily mitigate the effects of ischemia, leading to stabilization or even a slight decrease in neuronal firing rates.

Chemogenetic inactivation of forebrain excitatory neurons in transgenic mice has been used to investigate the contribution of those neurons to the acute development of focal ischemia. ¹⁷ However, broad suppression of excitatory neurons across the whole forebrain is not specific enough to target the roles of individual structures. Here, to expand on the role of the hippocampus in global cerebral ischemia and to assess the neuroprotective effect of chemogenetic manipulation, we precisely and selectively modulated hippocampal excitatory neurons at the acute stage of the global cerebral ischemia model. We combined this with detailed electrophysiological measurements to thoroughly assess the role of neuronal activity in the development of brain damage.

Astrocyte reactivation is a major determinant of the outcome after cerebral ischemia. ⁶¹ , ⁶² In our study, chemogenetic attenuation of neuronal activity regulated astrocyte reactivity and, in particular, decreased the formation of astrocytes with the A1 neurotoxic phenotype. This suggests that reducing A1‐like astrocyte neurotoxicity after cerebral ischemia may be one of the mechanisms by which modulation of neuronal activity alleviate brain injury. Neurotoxic A1 astrocytes are thought to be induced through close cooperation with activated microglia. ⁶³ We speculate that neuronal activity may cause microglia activation, which in turn drives astrocyte transition. Future studies will be necessary to better understand the regulation of complicated cellular crosstalk by neuronal activity and to clarify the role of such crosstalk in ischemia‐induced injuries.

We recognize several limitations in the current study. First, increasing animal research has documented sex influences on brain anatomy, chemistry, and function. ⁶⁴ , ⁶⁵ Therefore, it is a limitation that should not be neglected that our animal models predominantly focus on male subjects. Including female subjects in preclinical research would provide more comprehensive insights. Second, we acknowledge that our findings currently provide a descriptive account of the phenomenon at hand. Indeed, the changes in neuronal activity following ischemia are highly dynamic and complex, requiring further experimental validation. Third, our current study did not specifically measure sleep time and exploratory behavior in sham versus ischemic animals. This oversight will be addressed in future research to offer a more comprehensive assessment of the impact of experimental conditions on animal behavior.

Conclusions

In conclusion, our study systematically uncovers the time course of electrophysiological changes in CA1 hippocampus in awake rats during the acute stages of ischemia and reperfusion, especially changes in single‐unit firing and LFPs. The finding that chemogenetic inhibition of excitatory CA1 neurons relieves ischemic damage indicates that neuronal hyperexcitability is a pivotal factor for hippocampal injury in the progression of global cerebral ischemia and suggests that modulation of neural activity may have value in the development of novel therapies for ischemic brain injury. Then, our findings might pave the way for the promising therapies targeting brain activity and neural networks in the treatment of brain ischemia and help to optimize manipulation paradigms.

Sources of Funding

This work was supported by the National Natural Science Foundation of China (32271055 and 32070995 to Anan Li) and by the Jiangsu Province Innovative and Entrepreneurial Team Program to Anan Li.

Disclosures

None.