---
title: "Spreading depolarization alters distant cortical connectivity after focal injury"
id: "plos-one-10-spreading-depolarization-drives-distant-connectivity-changes-in-the-focally"
canonical_url: "https://medichelpline.com/clinical-feed/plos-one-10-spreading-depolarization-drives-distant-connectivity-changes-in-the-focally"
content_type: "clinical_feed_article"
specialty: "Neurology"
source_name: "PLOS ONE (Medicine)"
source_url: "https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0362090"
published_at: "2026-11-03T14:00:00.000Z"
evidence_level: "Journal Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Spreading depolarization alters distant cortical connectivity after focal injury
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/plos-one-10-spreading-depolarization-drives-distant-connectivity-changes-in-the-focally
- **Specialty:** [Neurology](https://medichelpline.com/clinical-feed/neurology.md)
- **Primary Source:** PLOS ONE (Medicine)
- **Source URL:** [Original Journal Publication](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0362090)
- **Published At:** 2026-11-03T14:00:00.000Z
- **Evidence Rating:** Journal Feed
## Executive GIST (TL;DR)
- The investigators recorded local field potentials (LFPs) in freely behaving rats after a focal frontal micro-injury and observed propagating cortical **spreading depolarization** (SD) confined to the ipsilesional hemisphere. - SD produced a transient increase in within-hemisphere synchrony (hypercoupling) during the depolarization wave, followed by prolonged interhemispheric decoupling after SD termination. - Directional connectivity analysis (transfer entropy) indicated increased influence from the uninjured hemisphere onto the injured hemisphere during the post-SD hypoconnectivity period in both somatosensory and retrosplenial cortices. - The retrosplenial cortex, a hub of the **default-mode network**, started with strong interhemispheric coupling at baseline and exhibited complex, region-specific connectivity changes after SD. - Functional connectivity (non-directional) was estimated using the **mutual information** function; effective (directional) connectivity used **transfer entropy**. Both measures were computed in 20-second nonoverlapping epochs. - Experimental dataset derived from seven implanted Wistar rats; one animal excluded for improper cannula placement and six artifact-free LFP recordings (L = 6) entered the connectivity analysis. - Microinjury was delivered through implanted guide cannulas in awake animals; recordings covered 400 s after SD initiation and a separate baseline epoch (800–1100 s) for control values. - The source report did not provide exact stereotaxic coordinates for some implant targets where placeholders appear, and it did not report pharmacologic manipulations or behavioral cognitive testing tied to connectivity changes. - Investigators deposited raw data in a public Zenodo repository and acknowledged funding support from the Russian Science Foundation; no competing interests were declared.
## Clinical Analysis & Structured Key Points
Spreading depolarization drives distant connectivity changes in the focally injured cortex | PLOS One Browse Subject Areas ? Click through the PLOS taxonomy to find articles in your field. For more information about PLOS Subject Areas, click here . Article Authors Metrics Comments Media Coverage Reader Comments Figures Figures Abstract Cortical spreading depolarization (SD), a slow-propagating wave of transient cellular depolarization, reliably occurs after acute brain injury but usually remains undetected. Focal brain injury is known to produce a network-level dysfunction that changes the strength and direction of connectivity far beyond lesion site (connectional diaschisis) that can underlie post-injury neurological dysfunction. We suggested that injury-induced SD could contribute to connectional diaschisis and studied the effects of SD on functional and effective connectivity in neural networks supporting cognitive and sensory functions. Electrographic recordings of cortical activity were obtained in freely behaving rats immediately after focal micro-injury of the frontal cortex. The injury-induced SD propagated over the cortex of the ipsilesional hemisphere but never invaded the contralesional cortex. Strength and direction of connectivity were estimated using non-directional (mutual information function) and directional (transfer entropy) measures calculated for recordings from the retrosplenial (a key hub of the default-mode network) and somatosensory cortex. Mutual information analysis showed initial hypercoupling within the injured hemisphere (during SD) followed by longer interhemispheric decoupling (after SD termination). During post-SD hypoconnectivity, influence of the uninjured cortex on the injured one increased in both cortical regions. The retrosplenial cortex, component of the default-mode network, showed strong interhemispheric coupling during the baseline period and complex changes in connectivity after SD. Therefore, SD arising in the injured cortex disturbs long-range connectivity in distant cortical regions and causes widespread network alterations involving the default-mode network. Given the key role of the network in cognition, connectivity changes produced by injury-induced SD could contribute to the mechanisms of cognitive deficit after brain injury. Citation: Lachinova DA, Smirnova MP, Sysoev IV, Vinogradova LV (2026) Spreading depolarization drives distant connectivity changes in the focally injured cortex. PLoS One 21(11): e0362090. https://doi.org/10.1371/journal.pone.0362090 Editor: Giuseppe Biagini, University of Modena and Reggio Emilia, ITALY Received: June 22, 2026; Accepted: October 20, 2026; Published: November 3, 2026 Copyright: © 2026 Lachinova et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: All relevant data for this study are publicly available from the Zenodo repository ( https://doi.org/10.5281/zenodo.22898499 ). Funding: This work was supported by Russian Science Foundation, Grant No. 22-15-00327-P The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declare that no competing interests exist. Introduction Cortical spreading depolarization (SD) is a wave of intense cellular depolarization propagating over the cortex with low (3–5 mm/min) velocity and causing transient changes in neuronal activity, metabolism and blood flow in the affected tissue [ 1 – 3 ]. Clinical studies have shown that SD is a reliable component of the cortical response to acute brain injury but conventional scalp EEG is unable to detect SD. Therefore, in most clinical situations the occurrence of SD remains unrecognized. In patients with stroke or brain trauma, intracranial recordings detected clusters of recurrent cortical SDs during acute and subacute periods [ 4 , 5 ]. Because the clusters can contribute to secondary brain injury, cortical SD is currently recognized as an important target for monitoring and therapy in patients with acute brain insults [ 5 ]. Focal brain injury is known to produce functional changes far beyond the lesion site (diaschisis). The distant changes occur within minutes after acute insults and gradually regress over time [ 6 , 7 ]. Connectional diaschisis, a network-level dysfunction observed after stroke or traumatic brain injury, includes changes in patterns, strength and direction of connectivity in intact areas far from injury [ 6 ]. Functional imaging studies have shown that a local cortical damage produces widespread changes in connectivity in both hemispheres and between them [ 8 ]. In patients with brain injury, connectivity changes in the motor cortex are thought to contribute to motor impairment and normalization of connectivity correlates with recovery of motor function [ 8 , 9 ]. Now, connectivity analysis is suggested to provide new information on the pathophysiological mechanisms of acute brain injury, its consequences and therapy [ 6 ]. Being locally induced, cortical SD reliably invades distant regions of the ipsilateral hemisphere [ 1 , 2 ]. Therefore, SD waves that arise in the injured brain could contribute to diaschisis, including the connectional one. Previously, we have found that the propagation of SD to unimodal (motor or sensory) cortical regions transiently disrupts interhemispheric functional connectivity [ 10 – 12 ]. However, it is unknown whether SD affects connectivity in transmodal cortical networks, particularly in the default-mode network (DMN). The widely distributed network increases its activity at rest and plays a crucial role in internally-directed cognition. Abnormal patterns of DMN connectivity have been found in patients with traumatic brain injury, schizophrenia, autism, depression [ 13 , 14 ]. Resting-state functional connectivity in the DMN is extensively studied using a functional neuroimaging approach and DMN connectivity measures are suggested as a biomarker of cognitive state for clinical conditions [ 14 , 15 ]. Disconnectivity within the DMN network is supposed to contribute to cognitive deficit after acute brain insults [ 16 ]. We hypothesized that SD triggered by an acute focal injury of the cortex could contribute to connectional diaschisis and mediate distant connectivity changes in cortical networks that support cognitive and sensory functions. To examine the suggestion, we studied the effect of SD developed in response to a distant focal injury of the cortex on functional and effective connectivity in intact cortical regions, the retrosplenial and somatosensory cortices. The retrosplenial cortex (posterior cingulate cortex in humans) is a key hub of the default-mode network and transmodal association cortex that supports high cognitive functions [ 13 , 17 ]. The somatosensory cortex is a unimodal sensory cortex that plays a key role in processing sensory information. Changes in strength and direction of intracortical (interhemispheric and intrahemispheric) connections were estimated using non-directional (the mutual information function) and directional (transfer entropy) measures. Given that electrophysiological-based approach provides direct assessment of functional connectivity with high temporal resolution, we calculated the connectivity measures using electrographic recordings of cortical activity obtained in freely behaving animals immediately after remote micro-injury of the cortex. Materials and methods Animals All experimental procedures were designed according to ARRIVE guidelines and the study was approved by the Ethics Committee of the Institute of Higher Nervous Activity and Neurophysiology (protocol No. 1, 11.02.2022). Wistar rats (3–4 months, males, 285–400) were received from Stolbovaya Animal Breeding Center (Moscow, Russia). Rats were housed in the animal facility in individual cages under controlled environmental conditions (a 12-h light-dark cycle, lights on at 7:00 A. M., 20–23°C) with free access to food and water. Well-trained and experienced stuff cared for animals and monitored their behavior and health every day. Efforts were made to minimize the number of used animals and their suffering. Rats were allowed to habituate for 2 weeks prior to electrode implantation surgery. Surgery Guide cannula for producing focal cortical micro-injury and electrodes for recording local field potentials (LFP) were stereotactically implanted under isoflurane anesthesia two weeks before the onset of experiments. Stainless-steel guide cannulas (outer diameter 0.6 mm; inner diameter 0.4 mm) aimed at middle layers of the frontal cortex (AP ; ML ) of the left and right hemispheres. The same length stylet was inserted into the guide cannula to prevent its obstruction. To record LFP, electrodes made of 0.25-mm-diameter teflon-coated silver wire (World Precision Instruments) were implanted in homotopic sites of the somatosensory (AP ; ML ) and retrosplenial (AP ; ML ) cortex of the two hemispheres. The reference electrode was placed over the cerebellum. Electrodes and guide cannulas were soldered to a pin connector and were fixed on the skull with acrylic cement. Experiment was performed 2 weeks after the surgery. During the post-surgery period, rats were injected with antibiotic amoxicillin (0.3 ml, i. m.) and glucose solution (1 ml, s. c.). All seven implanted rats were used in experiments; none died during the post-surgery period. Experimental design Experiments were performed in awake freely behaving rats with simultaneous LFP recording and video-monitoring of behavior. Each rat was individually placed in a shielded experimental chamber ( cm) and the implanted connector was attached to the recording cable. Electrical activity of the cortex (in a frequency band from 0 to 45 Hz) was recorded with a four-channel, high-input impedance (1gΩ) DC amplifier and A/D converter (E14-440, L-Card, Russia). The data were stored (1 kHz sampling rate, i. e. ms sampling interval) and analyzed off-line. After a 5-min period of habituation in the chamber and a 15-min recording baseline cortical activity, a microinjury of the frontal cortex of one hemisphere was performed and 15 minutes later identical injury of the symmetrical site in another hemisphere was produced. Microinjury was induced by a method described previously [ 18 , 19 ]. Briefly, the needle was introduced into cortical parenchyma via the implanted guide cannula and extended 1.0 mm from its tip. Animals were awake during the microinjury procedure and afterward, exhibiting locomotor or grooming behaviors. After the end of the experiment, rats were returned to their home cages; three days later all seven animals were euthanized using an overdose of anesthetics. None animal showed endpoint criteria for euthanasia (weight changes, abnormal behavior, neurological deficit etc.) or died before the end of the experiment. Recordings of experimental signals of 400 s length from the somatosensory and retrosplenial cortices of both hemispheres were obtained before and after focal cortical injury of each hemisphere. Occurrence of SD was defined by characteristic high-amplitude negative potential shift. Out of 7 rats with implanted electrodes, one rat was excluded from analysis due to incorrect localization of guide cannulas. Six artefact-free LFP recordings were selected for the connectivity analysis ( L = 6). Data preprocessing In each channel, the experimental time series of local field potentials v ( t ) was characterized by a slow trend of the mean of v ( t ) and occasional artifacts. We removed the trend by extracting the mean (1) in the moving time window with the window semilength T = 1 s: (1) (2) where is a value measured in the time moment , is a trend in this time moment (moving average) and is a corresponding detrended value, . By means of this numerical procedure, the signal was divided into two components: the detrended (fast component) signal U ( t ) which was further analyzed and the trend V ( t ) (slow component) which included a negative DC potential shift associated with SD. Then, high amplitude artifacts were removed from U ( t ). Some of them were signal transitions beyond the ADC (analog t o digital converter) dynamical range, so it was only necessary t o add 10 or mV to the signal. Recordings with more complex and long-lasting artifacts were excluded from the analysis completely or partly. Methods for connectivity reconstruction Here, we applied two measures for functional connectivity estimation. The first one is a mutual information function MI , which is a very popular and relatively simple nonlinear measure of signal similarity described, e.g., by [ 20 ] and used in neuroscience for at least 40 years [ 21 , 22 ]. There are different estimators for calculation of MI from experimental data (time series). We used the estimator proposed by [ 23 ] and based on [ 24 ] entropy since it demands a few data, being successfully applied in neuroscience in the moving window when studying other disorders like epilepsy [ 25 , 26 ]. It has also been recently applied to estimation of intracortical connectivity dynamics associated with SD [ 10 , 11 ]. However, MI cannot reveal the connectivity direction and measures only signal similarity or synchrony. To reveal the connectivity direction, we used another measure from information theory — transfer entropy TE [ 27 ]. The transfer entropy can also be considered as a version of nonlinear Granger causality [ 28 ] (the main measure of Granger causality is often referred as prediction improvement — PI ), and there are direct analytical proofs that for some very simple signals there are explicit formulae to calculate TE from PI and vice verse, but for complex signals this is not the case, see [ 29 ] for instance. Both measures were calculated in 20-s sequential nonoverlapping time epochs. For the control series the whole recording (from 800 to 1100 s) was used to calculate the baseline values of TE and MI . When analyzing the response to SD, 400 s of recording were used, obtained directly after wave initiation ( K = 20 epochs). For each i -th ( ) time epoch for each recording the values of MI and TE were calculated in four channel pairs, given four values of MI and eight values of TE (due to directionality for each channel pair there could be two values of TE and the only value of MI since this measure is indirectional). Next, MI and TE were calculated for two intrahemispheric channel pairs (one corresponding to the affected hemisphere and the other corresponding to nonaffected one) for the symmetric channels: in the somatosensory cortex: , and and in the retrosplenial cortex: , and . Then, MI and TE were calculated for two intrahemispheric channel pairs: in the affected hemisphere: , and and in the nonaffected one: , and . Here and further, the notation is as follows: a stays for the affected hemisphere and n — for the nonaffected one, s — for the somatosensory cortex and r — for the retrosplenial cortex. Mutual information function Mutual information function MI (3) can be considered as a generalization of correlation function, which takes into account all types of signal inter-dependencies: not only linear as correlation does, but also nonlinear. (3) where H ( X , Y ) is the joint entropy of signals X and Y and H ( X | Y ) and H ( Y | X ) are conditional entropies. If the entropies were calculated by mean of estimation of probability density from bins (i. e. by calculating the probabilities, and that the corresponding value belongs to i -th bin by X and j -th bin by Y ), the calculation formula would be (4). (4) The formula (4) is nonparametric and includes all possible nonlinear interdependences, but its application is problematic due to bad statistical properties of bin-based estimates: large errors for significantly nonuniform distributions together with demanding too many data for accurate estimation. We have to say that the differential entropy still has troublesome properties regardless the way it is calculated [ 30 ]. Difficulties in calculating MI forced scientists to use different parametric nonlinear correlation coefficients instead, see the popular approach by [ 31 ] for instance. Such coefficients may work fine for some dependencies and poorly for others due to different actual nonlinearities, which are unknown in practice, but have to be explicitly approximated. Actually, the problem of MI calculation from short series (or samplings) was mostly solved in [ 24 ] with introducing a new estimator based on nearest neighbors instead of direct use of bins, but this work had bin unknown in neuroscience until the calculation algorithm was reformulated in [ 23 ]. Tests and comparisons showed this approach to be the best among others both by precision and computation complexity, see [ 32 ]. Since the problem of MI estimation is now mostly solved, different nonlinear correlation coefficients are no more relevant and we prefer direct use the algorithm by [ 23 ]. To calculate MI following [ 24 ] and [ 23 ] one should introduce the distance between data points and as a maximum of absolute values (5) rather than by means of Euclidean norm: (5) Then, for each data point the k -th nearest neighbor is detected using (4) and numbers of points which are ether X -neighbors of (this number is denoted as ) or Y -neighbors of (this number is denoted as ) are calculated. Then, the specific approximation of Shannon’s entropy using the distance in terms of (5) proposed by [ 24 ] gives us the formula (6): (6) where N is the number of data points and is a digamma-function. Estimates of MI ignore all continuous linear signal transformations, including signal shift and scaling. The only method parameter is k . Usually, k > 1 is recommended. We tested our results for stability using k = 1, k = 6 and k = 10 and found no significant difference. So, we used k = 1 for all cases for calculation simplify. Transfer entropy Transfer entropy was proposed in [ 27 ] as an alternative realization of the same concept as the nonlinear Granger causality [ 33 ]. It has been actively used in neuroscience for last decades [ 34 ]. The difference between the Granger causality and the transfer entropy is in using conditional entropies instead of forecasting models. Let us explain TE for a simple case of scalar signals. First, let us decide that we are interested to know whether the future of the signal from the system X measured as time series is determined not only by its current value, but also by the current value of a signal of another system Y measured as time series . So, let us denote the time interval between future time moment (for which we are interested in causality) and the current time moment as and let us denote for further simplicity. The conditional entropy H ( z | x ) describes the amount of information about the future of the signal which cannot be found in its current state. And the conditional entropy H ( z | x , y ) describes how much information we obtain measuring future values of in addition to what we have can get both from signal and signal at the current moment. If these two entropies: H ( z | x , y ) and H ( z | x ) are significantly different (the only possible way is H ( z | x ) > H ( z | x , y )), then we may accept the hypothesis that Y drives X since it helps to decrease uncertainty in the future of X . (7) For ergodic signals one can estimate the entropies by averaging over time instead of averaging over different series. For the calculations, we used the similar nearest neighbor approach as for MI , which was proposed by different groups multiple times, see [ 35 – 37 ]. The calculation formula is (8). (8) where is the same as in the (6) and and are calculated as numbers of neighbors for both X and Y or both
## Related Clinical Research

- [EEG-derived indices classify sporadic Creutzfeldt-Jakob disease from Alzheimer’s and healthy aging](https://medichelpline.com/clinical-feed/plos-one-17-classification-of-sporadic-creutzfeldt-jakob-disease-based-on-resting-state.md)
- [Community Support Intervention for Driving Cessation Post-Acquired Brain Injury in Japan](https://medichelpline.com/clinical-feed/plos-one-8-study-protocol-for-a-non-randomized-trial-of-community-participation-support.md)
- [High-resolution map of the basal ganglia reveals cell types tied to Parkinson’s and schizophrenia](https://medichelpline.com/clinical-feed/stat-news-1-stat-scientists-debut-map-of-basal-ganglia-a-key-brain-region-involved-in-an.md)
- [Neurological involvement in localized scleroderma: clinical features and neuroimaging patterns](https://medichelpline.com/clinical-feed/frontiers-in-immunology-9-neurological-involvement-in-localized-scleroderma-a-systematic-review-of.md)
- [Predicted Brain Age Differences Across Nine Disorders: Case-Control MRI Study](https://medichelpline.com/clinical-feed/plos-medicine-1-brain-aging-patterns-among-nine-neurological-disorders-a-case-control-study.md)

## Navigation
- [← Back to Neurology Feed](https://medichelpline.com/clinical-feed/neurology.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.