Cortical stroke affects activity and stability of theta/delta states in remote hippocampal regions

Abstract

Cognitive impairment is a common outcome of ischemic stroke. Our previous work has shown that an experimental stroke in the cortex reduces activity in remote hippocampal layers in rats. This study seeks to uncover the underlying functional connections between these areas by analyzing changes to oscillatory activity, signal power, and communication. We induced an ischemic stroke in the left somatosensory cortex of rats and used linear micro-electrode arrays to simultaneously record from cortex and hippocampus under urethane anesthesia at two weeks and one month after stroke. We found significant increase in signal power, as well as an increase in the number of brain state changes in response to stroke. Our results suggest that the cortex modulates the activity and stability of hippocampal oscillations, which is disrupted following cortical stroke that can lead to cognitive impairment.

I. Introduction

Stroke is the leading cause of long-term disability and affects approximately 800,000 people in the United states. In addition to the high-level processes which are disrupted by stroke, it is common to observe cognitive impairment. The hippocampus is well known for its role in learning and memory consolidation, however strokes induced by occlusion of the middle cerebral artery (MCA) rarely cause direct ischemic lesions in the hippocampal and parahippocampal areas [1,2]. Although cognitive impairment in the absence of hippocampal injury has been reported by a number of previous studies [3,4,5,6,7], the effect of ischemic stroke in the cortex on functional connectivity between cortex and hippocampus is not well understood.

Our study seeks to find deeper understanding of the relationship between the oscillations within the cortex and hippocampus after a chronic stroke. We performed experimental middle cerebral artery occlusion (MCAO) strokes in the somatosensory cortex [8,9] and recorded simultaneously from cortical and hippocampal layers at two weeks and one-month post stroke to compare the functional dynamics between these remote areas.

II. Materials and Methods

A. Animals

We conducted all experiments in accordance with the animal care guidelines issued by the National Institutes of Health and by the San Francisco VA Medical Center Institutional Animal Care and Use Committee. Adult male Sprague-Dawley rats approximately 2.5 months of age weighing 250g (Charles River Laboratories, Wilmington, MA) were used.

B. Experimental Stroke

We unilaterally induced stroke in rats with the MCAO method using isoflurane (1.5%) / O2 (30%) / N2O (68.5%) anesthesia as described previously [10], producing cortical infarct restricted to the somatosensory cortex [11]. In brief, we made a 1.5 mm diameter burr hole 1 mm rostral to the anterior junction of the zygoma and temporalis bone with a dental drill. The dura mater was pierced with a 30-gauge needle, and the main trunk of the left MCA was ligated permanently above the rhinal fissure, resulting in ischemic damage in the left hemisphere. Sham group rats received the same treatment except they did not receive occlusion of the MCA. Control rats did not undergo any surgery.

C. Recording

We used multisite extracellular silicon probes (NeuroNexus Technologies) to produce electrophysiological recordings under urethane anesthesia for approximately two hours (Sigma, 15 mg/kg i.p.) as described previously [11,12]. We implanted two electrodes to allow recoding from cortex and hippocampus of both ipsilateral and contralateral hemispheres. Following craniotomy, electrodes were inserted into the brain after the dura mater was resected to target the dorsal hippocampus at [AP: −3.3 mm; ML: +/− 2 mm] via a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA) (Fig. 1). Proper recording locations were identified by characteristic signals from the CA1 pyramidal, radiatum, and molecular layers.

Figure 1.

Schematic of infarct area and probe locations. Probes are inserted to cover sensory motor cortex and hippocampus. Approximate infarct and peri-infarct areas from stroke demarked by grey shading.

D. Data preprocessing

We stored the data at 32K Hz after band-pass filtering (0.1-9K Hz) with an input range of ± 3 mV (Digital Lynx SX, Neuralynx, USA). Prior to data processing, the precise locations of the pyramidal and radiatum layers were confirmed manually by using off-line analysis software (Neuroscope, GNU). All recordings were down-sampled to 1250 Hz (Matlab, MathWorks, USA) for analysis.

E. Data processing

We used the local field potentials (LFP) of four channels (cortex, CA1 oriens, stratum lacunosum- moleculare (SLM), and pyramidal) in our analysis. Individual brain waves were isolated from the LFPs by band-pass filtering the following frequency ranges: delta (0.1-3Hz), theta (4-7Hz), alpha (7-13Hz), beta (13 - 30Hz), gamma (30 - 58Hz), and high-gamma (62 - 200Hz). We calculated the signal power of each frequency band using smoothed Hilbert amplitudes for each layer.

We defined two brain states in the anesthetized mouse analogous to REM and non-REM states during sleep by calculating the ratio of theta to delta band amplitudes (TD) (Fig. 2) using the amplitude envelope of Hilbert transform for delta and theta from the SLM layer [13,14,15]. The envelope was smoothed with a Gaussian filter with a width of 300ms. Two brain states were defined for this analysis, high theta/delta (HTD) and low theta/delta (LTD), which were calculated using the ratio of theta to delta for each LFP (Fig. 2). We set the HTD threshold value separately for each animal to prevent noise and small perturbations from reporting excessive state-changes.

Figure 2.

Detecting high and low TD states. The columns show examples of LFP traces for control and stroke. The rows are: 1- Band-pass filtered theta, 2- Band-pass filtered delta, 3- Hilbert amplitude of Theta, 4- Hilbert amplitude of delta, 5- Theta after Gaussian smoothing. 6- Delta after Gaussian smoothing, 7- Ratio of theta / delta, 8- The theta / delta ratio after Gaussian smoothing. Grey areas represent HTD period.

III. Results

A. Signal power is affected in both cortex and hippocampus following stroke

To quantify the change in brain activity during chronic stage of stroke, we calculated the signal power of each frequency band in each layer and compared the ipsilateral hemisphere to the contralateral hemisphere (Fig. 3).

Figure 3.

Comparison of high-gamma signal power between ipsilateral and contralateral hemispheres. Significant differences (P < 0.01) are demarked with two asterisks, (P < 0.001 are demarked with three asterisks)

In the cortex, CA1 SLM and oriens layers, all frequency bands showed a significant increase in signal power in the ipsilateral hemisphere (P < 0.001) in comparison to the contralateral side following stroke (Fig. 3). In CA1 pyramidal layer we found delta, alpha, beta, and high-gamma bands to significantly increase in signal power in the ipsilateral hemisphere (P < 0.007) following stroke. We did not detect any significant difference in brain activity between hemispheres in either stroke or sham groups (P > 0.19). We also detect a smaller increase in power in animals recorded 2 weeks (2WK) after stroke compared to those recorded one-month (1M) stroke (P < 0.001).

B. Hippocampal TD states change at a higher rate following stroke

To quantify the amount of state changes occurring in stroke and non-stroke groups, we compared stroke to control and sham, grouping ipsilateral and contralateral hemispheres separately (Fig. 4). In both hemispheres, we saw a significant increase in the state change in stroke groups in comparison to control (P < 0.05) while there was no significant change between sham and control (P > 0.57) (Fig. 4).

Figure 4.

Average number of state changes. Ipsilateral and contralateral hemispheres compared separately. Significant differences (P <0.05) demarked with an asterisk.

IV. Discussion

Previous work has shown that Stroke can lead to changes in functional connectivity between the cortex and hippocampus [16,17,18], and that MCAO stroke-affected mice display fewer activated neurons in the hippocampus. Stroke has also been shown to cause changes in brain oscillations [19,20,21] Additionally, it has been shown that theta activity in the hippocampus is correlated with spatial memory and memory consolidation during REM stages in sleep [22,23]. Further, manipulating theta activity changes cognition, supporting theta’s causal effect on cognition [24,25].

Interestingly, we found the increase in signal power observed in the cortex mirrored by a similar increase in signal power in remote hippocampal areas across multiple frequency bands. This indicates that the increased activity of the cortex may modulate the activity of the hippocampal layers. Significant changes are more prevalent one month after stroke than two weeks after stroke, which may indicate some form of compensatory mechanism induced by the hippocampus. High-gamma bands were the most consistently affected by cortical stroke, which may indicate that these oscillations are responsible for the modulation of the hippocampus (Fig. 3).

TD ratios are regarded as a potential biomarker for post-stroke cognition [26] because they are often associated with reduced neurological function [27,28]. We found that rate that high and low TD states alternated in the stroke group significantly increased compared to control group in both ipsilateral and contralateral hemispheres, which indicates that the lesion in the left somatosensory cortex is affecting the network dynamics on both ipsilateral and contralateral sides of the brain.

In summary, the changes in power in damaged areas and remote areas may be a biomarker of functionally connected brain regions. The frequency of state changes alternations may be another useful biomarker in disrupted networks, however deeper understanding is needed to uncover the underlying mechanisms which control the rate of state change and may contribute to a deeper understanding of how memories are encoded and how spatial memory is affected by this disruption.