Subarachnoid hemorrhage mediates human neocortical network, membrane potential, and action potential bursting via glutamate receptors

Abstract

Subarachnoid hemorrhage (SAH) is a life-threatening neurological emergency with high mortality and morbidity rates. Despite its severity, the acute and direct neuronal effects of SAH remain poorly understood, particularly in human brain tissue. To address this gap, we applied cerebrospinal fluid from individuals with SAH (SAH-CSF) to human neocortical slices and examined neuronal activity and intrinsic properties. We found that SAH-CSF significantly increased population-wide neuronal activity, depolarized membrane potential, and either elevated firing rates or induced depolarization block of action potentials in these human neocortical slices. Notably, these neuronal changes were reversible upon washout of the SAH-CSF. Furthermore, kynurenic acid (KYNA), a glutamate receptor antagonist, effectively prevented SAH-CSF-induced neuronal changes. Together, these findings revealed key neuronal consequences of SAH-CSF in human brain tissue and suggest that glutamate receptor antagonists may offer therapeutic potential for SAH.

Acute and direct neuronal effects of subarachnoid hemorrhage mediated by glutamate receptors reveal therapeutic potential.

Introduction

Subarachnoid hemorrhage (SAH) is a life-threatening neurological disorder characterized by high mortality and morbidity rates^1–3. Moreover, SAH acutely induces spreading depolarizations and early cortical infarction^4–8, impairs blood-brain barrier integrity⁹ and increases seizure activity^9,10. Despite past advancements in its acute management and treatment, the improvement of SAH-related outcomes has recently plateaued¹¹.

Several mechanisms underlying SAH have been elucidated through studies in both human patients and murine models. For instance, key metabolomic and lipidomic pathways have been identified in individuals with SAH¹², alongside spreading depolarization, and disturbances in the vascular and immune system^13,14. Murine models of SAH have been instrumental in investigating potential therapeutic interventions and exploring the molecular and cellular mechanisms involved^15–27. Despite these advancements, the direct and acute neuronal consequences of SAH remain poorly understood, particularly in human brain tissue. Investigating human brain tissues offers the potential to provide critical, direct evidence and insights into these pathogenic mechanisms.

To address this knowledge gap and fulfill an unmet clinical need, we first measured both population-level neuronal activity and cell-type-specific neuronal excitability and synaptic properties in human neocortical slices. We observed that cerebrospinal fluid (CSF) from individuals with SAH (SAH-CSF) significantly increased neuronal activity at the population level. Moreover, SAH-CSF depolarized neuronal membrane potential, and either increased firing rates or induced depolarization block of action potentials. Notably, SAH-CSF-induced neuronal changes were reversed following the washout of SAH-CSF. Furthermore, the use of glutamate receptor antagonists effectively prevented these changes. In summary, our findings provide insights into the acute and direct impacts of SAH and suggest that glutamate receptor antagonists may offer promising therapeutic benefits for individuals affected by this condition.

Results

SAH-CSF increased population-wide neuronal activity in human neocortical slices and anesthetized mice

Surgical resection remains the only feasible method for accessing live human brain tissue for ex vivo analyses^28–30. To investigate the acute and direct effects of SAH in human brain tissue, we obtained human neocortex samples from patients undergoing tumor excision surgery (Fig. 1a; and Supplementary Table S1). Human peritumoral neocortex was verified (Fig. 1b) and used for subsequent experiments. We then performed two-photon calcium imaging using the calcium indicator Cal-520 on these human neocortical slices (Fig. 1c). Three conditions were applied: artificial cerebrospinal fluid (aCSF), cerebrospinal fluid (CSF) from individuals with hydrocephalus (HC), and CSF from individuals with subarachnoid hemorrhage (SAH). We included CSF from individuals with HC as controls, in addition to aCSF, because human CSF itself can induce changes in neuronal excitability in resected human brain slices^31,32. Next, we analyzed 12 representative cells across these conditions (Fig. 1d). Under SAH conditions, most neurons exhibited increased activity (C01–C06, C08, C10–C12), while a few displayed decreased activity (C07, C09) compared with the aCSF and HC conditions (Fig. 1e, Supplementary Movies S1–S3). Overall, SAH-CSF significantly increased population-wide neuronal activity, peaking 15 min after treatment (Fig. 1f, g). In summary, two-photon calcium imaging revealed that SAH-CSF increased neuronal activity at the population level in human neocortical slices.

Fig. 1. SAH-CSF increases population-level neuronal activity in human neocortical slices.

a Schematic representation of human neocortex tissue resected during neurosurgical procedures for brain tumor removal (left), alongside a representative image of the resected neocortical slices (right). b Immunofluorescence staining of brain tumors and adjacent neocortical tissue using antibodies against astrocytes (GFAP antibody, red) and neurons (NeuN antibody, green). Nuclei were counterstained with DAPI. Scale bar = 50 μm. c Schematic illustrating two-photon imaging of human neocortical slices loaded with Cal-520. d Upper panels: Left, representative image of Cal-520-labeled cells in human neocortical slices under artificial cerebrospinal fluid (aCSF) conditions, captured via a two-photon microscope. Scale bar, 100 µm. Right, representative field of view highlighting regions of interest (ROIs) after cell extraction. Lower panel: Representative raw fluorescence traces from individual neurons under aCSF conditions. Scale bar: 60 sec, ΔF = 5. e Heatmap displaying neuronal activity of individual neurons from resected neocortical slices during sequential addition of different solutions. f Population-level neuronal activity plotted over time from human neocortical slices during continuous solution application. g Average neuronal activity within defined recording windows from resected neocortical slices. Data are presented as means ± s.d. or as box-plots showing the median and interquartile range (IQR), with the whiskers denoting the minimum and maximum values. Statistical analysis was performed using Two-way ANOVA followed by Scheffe post hoc comparison. *P  <  0.05, **P  <  0.01. N = 12 cells from one slice of a single subject. HC hydrocephalus, SAH subarachnoid hemorrhage, 3% SAH: 3 ul SAH-CSF + 97 ul aCSF; 5% HC: 5 ul HC-CSF + 95 ul aCSF.

Comparable results were obtained in the neocortex of anesthetized mice using in vivo two-photon calcium imaging with Cal-520 (Fig. 2a–c). Administration of 0.5% SAH-CSF produced a marked increase in fluorescence intensity compared with 0.5% HC-CSF treatment (Fig. 2d and Supplementary Movies S4–S6). This robust elevation in fluorescence intensity indicates that SAH-CSF rapidly and strongly activates neocortical neurons, elevating excitability and calcium dynamics in vivo. Collectively, these findings identify SAH-CSF as a potent, immediate stimulator of neuronal activity in both human and mouse neocortex.

Fig. 2. SAH-CSF increased population-wide neuronal activity in the neocortex of anesthetized mice.

a Schematic of experimental setup. Images for b injection setup and c external (left) and internal (right) views of the cranial window. d Average fluorescence ratios in recording window (∆F_{HC or SAH}/∆F_aCSF). Data are presented as box-plots showing the median and IQR, with the whiskers denoting the minimum and maximum values. Statistical analysis was performed using Wilcoxon matched-pairs signed rank test, ***P  <  0.001. N = 29 cells from three mouse brains (one male and two females). HC hydrocephalus, SAH subarachnoid hemorrhage. 0.5% SAH: 5 ul SAH-CSF + 995 ul aCSF; 0.5% HC: 5 ul HC-CSF + 995 ul aCSF.

Neuronal excitability and synaptic transmission were robustly measured in human neocortical slices and distinct from mouse counterpart

To further investigate how SAH-CSF increased neuronal activity, we first established a platform for whole-cell patch-clamp recordings on human neocortical slices (Fig. 3a, d, g). Additionally, to differentiate between excitatory and inhibitory neurons, we conducted post hoc staining with the GABAergic inhibitory neuron marker, GAD67 (Fig. 3b, c) and assessed electrophysiological properties. The firing patterns of action potential differed between excitatory and inhibitory neurons, with excitatory neurons exhibiting firing rate adaptation beyond a stimulus of 120 pA (Fig. 3e, f, top). Detailed action potential parameters were provided for both excitatory (Fig. 3e, bottom) and inhibitory neurons (Fig. 3f, bottom). Notably, the firing rates of excitatory neurons in human neocortex were significantly higher compared with mouse counterparts (Supplementary Fig. S1a).

Fig. 3. Neuronal and synaptic properties were reliably measured in excitatory and inhibitory neurons of human neocortical slices.

a, d, g Schematic illustration of whole-cell patch-clamp recordings performed on human neocortical slices. Representative images of excitatory neurons (b) and inhibitory neurons (c) from human neocortical slices. Excitatory and inhibitory neurons were distinguished via post hoc immunofluorescence staining using an anti-GAD67 antibody. Scale bar = 50 μm. Representative electrophysiological traces and corresponding quantification from excitatory (e) and inhibitory (f) neurons recorded in current-clamp mode. Scale bar = 50 μm. N = 9 excitatory neurons from nine slices of seven subjects; N = 25 inhibitory neurons from 25 slices of 13 subjects. Scale bar: 200 ms, 20 mV. Representative traces and quantitative data from excitatory (h) and inhibitory (i) neurons recorded in voltage-clamp mode. N = 9 excitatory neurons from nine slices of seven subjects; N = 9 inhibitory neurons from nine slices of nine subjects. Scale bar = 200 ms, 20 pA. Data are presented as means ± s.d. or as box-plots showing the median and IQR, with the whiskers indicating the minimum and maximum values. AP action potential, Rin input resistance.

Next, we recorded the frequency and amplitude of spontaneous excitatory postsynaptic currents (sEPSCs) from excitatory neurons (Fig. 3h) and inhibitory neurons (Fig. 3i). Interestingly, while the frequency of sEPSCs did not differ significantly, the amplitude of sEPSCs in human neocortical excitatory neurons was significantly higher compared with mouse counterparts (Supplementary Fig. S1b). In summary, we established a robust and reliable platform for measuring neuronal excitability and synaptic properties in human neocortical slices with cell-type-specific resolution.

SAH-CSF-induced changes in neuronal properties in human neocortical slices

With reliable whole-cell patch-clamp recordings established in human neocortical slices, we next investigated the effects of CSF from individuals with SAH. As controls, aCSF and CSF from individuals with HC were applied in the initial experimental conditions. Specifically, HC-CSF and aCSF were served as control and reference conditions respectively for the comparison. Subsequently, CSF from individuals with SAH was introduced, followed by a washout using aCSF (Fig. 4a, top; 4b). We observed that SAH-CSF significantly depolarized the neuronal membrane potential (Fig. 4c, d) and either elevated the firing rates (Fig. 4a(i)) or induced depolarization block (Fig. 4a(ii)) of action potentials in human neocortical slices. Importantly, the removal of SAH-CSF fully reversed these SAH-CSF-induced changes in neuronal properties (Fig. 4c, d). Moreover, we did not observe significant variation of SAH-CSF-induced changes in neuronal properties attributable to the collection days of post-SAH (Supplementary Fig. S2). In summary, our findings demonstrated that SAH-CSF depolarized neuronal membrane potential and altered action potential bursting in human neocortical slices.

Fig. 4. SAH-CSF depolarized membrane potentials and altered firing rates of action potentials in human neocortical slices.

a Schematic diagram outlining the experimental design. Two representative trace types (i, ii) were shown for four different conditions: aCSF (artificial cerebrospinal fluid), HC (hydrocephalus), SAH (subarachnoid hemorrhage), and washout with aCSF. Scale bar: 50 ms, 20 mV. b Schematic illustration of whole-cell patch-clamp recordings performed on human neocortical slices. c Membrane potential recordings from neurons under the four conditions. d Absolute values representing the difference in membrane potential between paired conditions. N = 9 cells (n = 5 excitatory neurons from five slices of five subjects and 4 inhibitory neurons from four slices of four subjects). Statistical analysis was performed using one-way repeated measures ANOVA followed by Dunn’s or Holm-Sidak’s post hoc comparison. **P  <  0.01, ***P < 0.001. Data are presented as means. Gray lines connect data from the same cells under different conditions. Vm: membrane potential.

Glutamate receptor antagonists fully prevented SAH-CSF-induced changes in neuronal properties in human neocortical slices

Membrane potentials are predominantly determined by ion concentration gradients across the membrane, and disruptions in ion homeostasis can impair normal physiology and contribute to diseases^33–35. Since SAH-CSF induced membrane depolarization in human neocortical slices, we first examined whether changes in ion concentrations in the CSF of individuals with SAH might underlie this effect. We measured potassium, sodium, chloride, and calcium levels, but found no significant differences compared with the CSF from individuals with HC (Fig. 5a). To further examine whether SAH-CSF caused depolarization via disturbing ion homoeostasis, we calculated the Goldman-Hodgkin-Katz (GHK) equation. The GHK equation is a well-established approach for estimating the resting membrane potential, as it incorporates the relative permeabilities of multiple ionic species across the neuronal membrane^36,37. Comparative analysis showed no statistically significant differences in equilibrium potentials between HC-CSF and SAH-CSF (Fig. 4b).

Fig. 5. Glutamate receptor antagonists fully prevent SAH-CSF-induced phenotypes in human neocortical slices.

a Concentrations of potassium, sodium, chloride, and calcium in HC and SAH conditions. N = 4 for individuals with HC. N = 4 for individuals with SAH. Statistical analysis was performed using Mann–Whitney test, two-tailed. b Membrane potentials calculated using the Goldman–Hodgkin–Katz (GHK) equation for aCSF, HC, and SAH conditions. c Schematic illustration of experimental design. Representative traces (i, ii) were shown for four conditions: aCSF, HC, SAH, and washout with aCSF, with all conditions supplemented with glutamate receptor antagonists (kynurenic acid (KYNA), 5 mM). Scale bar: 50 ms, 20 mV. d Schematic of experimental setup of cell patching on human neocortical slices. e Membrane potential recordings from neurons under the four conditions. f Absolute values representing the difference in membrane potential between two assigned conditions. N = 6 cells (n = 1 excitatory neuron from one slice of a single subject and 5 inhibitory neurons from five slices of two subjects). Data are presented as means ± s.d. Data are presented as means or as box-plots showing the median and IQR, with the whiskers denoting the minimum and maximum values. Gray lines connect data from the same cells under different conditions. Vm membrane potential.

Previous studies have reported elevated glutamate levels as a key metabolite in the CSF of individuals with SAH^38,39. To investigate the role of glutamate in SAH-CSF-induced changes in neuronal properties, we applied kynurenic acid (KYNA), a broad-spectrum glutamate receptor antagonist, to SAH-CSF-treated human neocortical slices (Fig. 5c, top; 5d). As controls, aCSF and CSF from individuals with HC, both supplemented with KYNA, were used in the initial experimental conditions. We found that the addition of 5 mM KYNA fully prevented the SAH-CSF-induced neuronal alterations (Fig. 5c(i), (ii), e, f). These findings suggest that SAH modifies neuronal properties primarily through the activation of glutamate receptors in human neocortical slices.

Discussion

The direct and acute effects of SAH are essential for understanding the mechanisms underlying early brain injuries and developing novel therapeutic interventions. Utilizing live human brain tissues provides direct evidence and valuable insights into these effects. Through assessing both population-level neuronal activity and cell-type-specific excitability and synaptic properties in human neocortical slices, we found that SAH-CSF increased population-wide neuronal activity, depolarized membrane potentials, and either increased firing rates or induced depolarization block of action potentials. Notably, the removal of SAH-CSF reversed these SAH-CSF-induced neuronal changes. Additionally, the application of glutamate receptor antagonists completely prevented these changes.

Calcium channel blockers are the primary medications for individuals with SAH given its neuroprotective properties⁴⁰. Although they do not directly influence early brain injury following SAH, they are administered because of their mild yet significant protective effect against delayed ischemia, likely through partial antagonism of spreading depolarization-induced spreading ischemia^41–43. Interestingly, our findings showed that applying glutamate receptor antagonists fully prevented the SAH-CSF-induced neuronal phenotypes. It would be valuable to further investigate whether NMDA receptor antagonists or AMPA receptor antagonists play a more significant role in mitigating brain injuries in rodent models of SAH. Notably, a previous study found that memantine, an NMDA receptor antagonist, alleviated neuronal injuries and improved neurological outcomes in a rodent model of SAH¹⁶. Similarly, NP10679, a GluN2B-selective NMDA receptor inhibitor, rescued neurological deficits in a rodent model of SAH¹⁵. Moreover, s-ketamine reduced the incidence of spreading depolarization in individuals with SAH⁴⁴. Together, these findings align with our results and underscore the therapeutic potential of targeting glutamate receptors in the treatment of SAH. Finally, ongoing clinical trials investigating NP10679 and ketamine warrant future attention and follow-up, despite previous clinical trials on NMDA receptor antagonists yielding discouraging results⁴⁵.

Two-photon calcium imaging revealed a peak in population-level neuronal activity 15 min after SAH-CSF application in human neocortical slices, followed by a mild decrease at 30 min (Fig. 1f, g). This mild reduction in activity may be attributed to glutamate reuptake by neighboring glia cells. Future studies could test this hypothesis by applying glutamate reuptake inhibitors to evaluate their effect on neuronal activity under SAH conditions.

CSD was first observed in rodent models of SAH⁴⁶ and was later detected in individuals with SAH^5,47. Moreover, CSD is strongly associated with increased early and delayed tissue injury and poorer clinical outcomes in individuals with SAH^47,48. Our findings demonstrate that SAH-CSF induces membrane potential depolarization, providing direct evidence to support this association. Furthermore, glutamate receptor antagonists completely prevented SAH-CSF-induced neuronal changes, suggesting their potential as an intervention for CSD.

One critical consequence of SAH is the disruption of blood-brain barriers (BBB)^49–52. BBB breakdown following SAH permits the extravasation of blood-derived components and changes in the composition of CSF including glutamate, potassium, hemoglobin, degradation products, and proinflammatory cytokines. These bioactive substances can modulate neuronal membrane properties, disturb astrocytic buffering, and impair inhibitory transmission, collectively leading to membrane depolarization and a lowering of seizure threshold. Our work extended previous findings and demonstrated that SAH-CSF directly and acutely depolarized membrane potential and either increased firing rates or induced depolarization block of action potentials in human neocortical slices in the absence of structural injuries. Additionally, observed SAH-CSF-induced neuronal phenotypes can be prevented from glutamate receptor antagonists. It would be of future interest to explore the impacts of other changed components of SAH-CSF on SAH-CSF-induced neuronal phenotypes.

Elevated extracellular potassium levels have been associated with poor outcomes in individuals with SAH^53,54, yet some studies have reported lower potassium concentrations in SAH-CSF compared with controls^55–57. In our study, we did not detect the differences in potassium levels between SAH-CSF and HC-CSF, possibly due to the limited sample size. That was one reason we focused on glutamate, a consistently elevated component of SAH-CSF, as a likely driver of the observed neuronal effects. It is also important to note that the extracellular potassium concentration within the clot on the brain surface is likely more relevant to the underlying cortical tissue than the potassium concentration in the CSF, which does not directly contact the cortex because it is covered by the clot⁵⁸. Indeed, the numerous cortical lesions following SAH typically occur at sites where clots lie directly on the cortical surface⁵⁹. Although investigating the direct effects of the clots on cortical tissue would be of greater physiological relevance than studying CSF alone, suitable experimental tools are still needed to address this question.

A comprehensive analysis of amino acid levels in SAH-CSF found that glutamate and other amino acids remained elevated at days 0–3, 5, and 10 post-SAH compared with controls³⁸. In our study, SAH-CSF was collected between days 1 and 6 post-SAH, a window that overlaps with this sustained glutamate elevation. We did not detect a clear influence of collection day on neuronal depolarization, but future studies could examine whether CSF collected at later time points produces reduced effects.

Several previous studies have demonstrated the feasibility of long-term cultures of human brain tissues^60,61. This opens up the possibility of manipulating neuronal circuits using optogenetic or chemogenetic approaches via adeno-associated viruses, as well as labelling specific cell types in the brain. Future studies could explore how SAH-CSF impacts neuronal circuits with a more detailed cell-type and pathway resolution in human neocortical slices.

In conclusion, our study highlights the direct and acute effects of SAH on human neocortical slices. Furthermore, our findings suggest that glutamate receptor antagonists may represent a promising acute therapeutic strategy for individuals affected by SAH.

Methods

Human brain tissue

Neocortex specimens were collected from patients undergoing brain tumor excision and have been described previously⁶². The neocortex was easily distinguished based on its morphology (Fig. 1a, right), and neocortical tissue adjacent to the tumors was used for subsequent experiments. Histochemistry analysis showed that brain tumor samples, such as glioblastoma, were enriched with astrocytes (GFAP-positive cells) but lacked neurons (NeuN-positive cells) (Fig. 1b, left). In contrast, adjacent neocortical samples were rich in neurons with minimal presence of astrocytes (Fig. 1b, right)^63,64. Detailed subject information is provided in Supplementary Table S1. The neocortical slices used in this study were obtained from a human population with a mean age of 60.24 years (SD = 13.43) and a sex distribution of 12 females and 9 males. All experimental protocols were approved by the Institutional Review Board (IRB) of National Taiwan University Hospital (IRB numbers: 202302097RIN) and conducted in accordance with the approved guidelines. Written informed consent was obtained from all participants.

Collection of human cerebrospinal fluid (CSF)

The collection method for human CSF has been previously described^26,27,65. Intrathecal CSF from individuals with SAH was obtained via lumbar puncture from the first to sixth day post-SAH. Detailed information was provided in Supplementary Table S2. The SAH-CSF used in this study was collected from a human population with a mean age of 61.57 years (SD = 11.09) and a sex distribution of 3 females and 4 males. For individuals with HC, CSF was collected during surgery through a ventricular catheter. The HC-CSF used in this study was obtained from a human population with a mean age of 74.44 years (SD = 6.54) and a sex distribution of 2 females and 7 males. All CSF samples were immediately centrifuged at 900 × g at 4 °C for 20 min to remove impurities including red blood cells. The supernatants were then divided into appropriate aliquots and rapidly frozen at –80 °C within 30 min of collection. Written informed consent was obtained from all participants or their legal representatives. All experimental protocols were reviewed and approved by the IRB of National Taiwan University Hospital (IRB numbers: 201301042RINB and 202112146RINA) and conducted in accordance with the approved guidelines.

Mice

C57BL/6J mice (Jackson Laboratories; stock no. 000664) were group-housed in ventilated cages with ad libitum access to food (PicoLab^® Rodent Diet 20, 5053) and water. Mice were maintained on a 12-h light/dark cycle (lights off at 8:00 pm). Mice used for electrophysiological recordings ranged from postnatal day 21–65, whereas those used for two-photon calcium imaging ranged from postnatal day 120–180. Electrophysiological recordings were conducted in the medial prefrontal cortex. Both male and female mice were utilized in the experiments. All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of National Taiwan University College of Medicine and the College of Public Health (IACUC No. 20220343, 20230047, and 20250068). The animal facility is accredited by AAALAC, and all experiments were conducted in accordance with approved guidelines.

Neocortical slice preparation

The methods for brain slice preparation have been previously described^63,66–69. Human neocortex was obtained via neurosurgical resection and immediately chilled in ice-cold dissection buffer containing (in mM): 87 NaCl, 2.5 KCl, 0.5 CaCl₂, 7 MgCl₂, 1.25 NaH₂PO₄, 25 NaHCO₃, 75 sucrose, 10 glucose and 1.3 ascorbic acid, oxygenated with 95% O₂ and 5% CO₂ (pH, 7.4; 300 mOsmol). Cortical slices (300 μm thick) were prepared in dissection buffer using a vibroslicer (VT 1200 S, Leica). Prior to recording, slices were incubated for 30 min at 30 °C in artificial cerebrospinal fluid (aCSF) consisting (in mM): 124 NaCl, 3 KCl, 2 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃ and 20 glucose, oxygenated with 95% O₂ and 5% CO₂ (pH, 7.4; 295 mOsmol). Mouse coronal neocortical slices were prepared following the same protocol as for human tissues.

Two-photon calcium imaging on human neocortical slices

Two-photon calcium imaging was performed using a two-photon microscope (MESOR01, mesoView) equipped with a water-immersion objective lens (Xlumplfln20xw, Olympus). Images were acquired at 1024 × 1024 pixels with a frame rate of 7.61 Hz, and a standard field of view of 0.5 mm × 0.5 mm. Data acquisition was controlled using mesoSoft + software (mesoView). Calcium imaging was facilitated by a Coherent Axon 920-1 TPC laser set to 920 nm with excitation power maintained at 60 mW under the objective to ensure consistent results without photobleaching. For the dye preparation, Cal-520^®, AM (ab171868, Abcam) was dissolved in dimethyl sulfoxide (DMSO) to create a 20 mM stock solution, then diluted in aCSF to a final concentration of 6 mM. The dye solution was incubated at 37 °C for one hour, followed by a solution refresh with new aCSF. Before imaging, samples underwent a 15-min dark adaptation period at room temperature. Calcium imaging experiments commenced immediately afterward.

Calcium image processing on human neocortical slices

Calcium imaging data were first concatenated using ImageJ software (https://imagej.nih.gov/ij/download.html). The concatenated videos were then processed using Inscopix Data Processing Software (IDPS, Inscopix). Pre-processing steps included motion correction and temporal down-sampling by a factor of second. Following pre-processing, candidate cells were identified and subjected to spatial filtering and temporal activity trace extraction. Individual neurons were manually identified and extracted based on their average Ca²⁺ transient waveforms and activity traces. The extracted data were normalized by calculating z-scores, where the mean activity of the entire recording was subtracted from each activity trace, and the result was divided by the standard deviation of each cell’s activity during the recording period. Neurons displaying activity peaks ≥ 3 standard deviations above baseline were classified as active.

Two-photon calcium imaging and image analysis in anesthetized mice

In vivo calcium imaging was performed in adult C57BL/6J mice with both genders (4–6 months old). Under aseptic conditions, mice were anesthetized with isoflurane (4–5% for induction, 1–3% for maintenance). A 2 × 2 mm cranial window was created over the neocortical area using a high-speed drill, and the dura mater was carefully removed to expose the neocortex. Cal-520^® AM (ab171868, Abcam) was prepared at a final concentration of 500 μM by dissolving the dye in DMSO and diluting with aCSF. The dye solution was slowly injected into the subcortical region at a depth of 0.5 mm using the multicell bolus loading technique with a 27-gauge needle. Imaging began 1.5 h after dye loading to allow sufficient diffusion and cellular uptake.

For calcium imaging, anesthesia was maintained with 0.5% isoflurane combined with carbogen gas, and a 3-min baseline fluorescence recording was obtained using aCSF perfusion. This was followed by administration of 0.5% HC-CSF for 5 min and a subsequent 3-min fluorescence recording. The cranial window adaptor was then rinsed with 2 ml aCSF before administration of 0.5% SAH-CSF for 5 min, followed again by a 3-min fluorescence recording. Fluorescence intensities were baseline-corrected and normalized to compare changes between treatments. Animals were euthanized after completion of the experimental protocol.

Whole-cell recordings

The methods for patch-clamp recordings have been previously described^63,66–70. For current-clamp recordings, aCSF was not supplemented with any synaptic blocker. The internal solution contained (in mM): 100 K-gluconate, 20 KCl, 0.2 EGTA, 10 HEPES, 4 ATP-Mg, 0.3 GTP-Na, and 0.025 Alexa Fluor^TM 594 Hydrazide (A10438, Thermo Fisher Scientific) (pH 7.2, 295 mOsmol). For voltage-clamp recordings, the internal solution consisted of (in mM): 100 CsCH₃SO₃, 15 CsCl, 2.5 MgCl₂, 10 HEPES, 5 QX-314∙Cl (L1663, Sigma), 5 BAPTA-TetraCs, 4 ATP-Mg, 0.3 GTP-Na, and 0.025 Alexa Fluor^TM 594 Hydrazide (A10438, Thermo Fisher Scientific)(pH 7.2, 295 mOsmol). To isolate spontaneous EPSCs (sEPSCs), 5 μM SR95531, and 1 μM strychnine were added to the aCSF and applied via a perfusion valve system (VC-8 valve controller, Warner Instruments, Hamden, CT, USA). To distinguish excitatory and inhibitory neurons, post hoc immunofluorescence staining with an anti-GAD67 antibody (1:500, MAB5406, Millipore) and analysis of action potential firing patterns were performed. The detailed staining protocols for anti-NeuN (1:500, MAB377, Millipore), -GFAP (1:500, G3893, Sigma), and -GAD67 antibodies have been described previously^63,64.

Cerebrospinal fluid (CSF) ion concentration analysis and the Goldman–Hodgkin–Katz (GHK) equation calculation

A 10 μl CSF sample was used to measure ion concentrations. The analysis was performed using the FUJI DRI-CHEM SLIDE chip Na-K-Cl/Ca in an automatic dry-chemistry analyzer (DRI-CHEM NX500i, Fujifilm), following the manufacturer’s instructions.

The GHK equation can be given by Eq. (1). For its calculation, we set the permeabilities of potassium (K⁺), sodium (Na⁺), and chloride (Cl⁻) ions in resting neurons were to 1.00, 0.05, and 0.45, respectively⁷¹.

$$V_m=\frac{RT}{F}\ln \left(\frac{p_K{\left[K^{+}\right]}_o+p_{Na}{\left[{Na}^{+}\right]}_o+p_{Cl}{\left[{Cl}^{-}\right]}_i}{p_K{\left[K^{+}\right]}_i+p_{Na}{\left[{Na}^{+}\right]}_i+p_{Cl}{\left[{Cl}^{-}\right]}_o}\right)$$ 1

Statistical analysis

Data are presented as means ± s.d. or as box-plots showing the median and interquartile range (IQR), with the whiskers denoting the minimum and maximum values. Sample sizes (n) are shown in the figures or specified in the text. Statistical analyses were conducted using SigmaPlot 13 (Systat Software). Normality was assessed using Shapiro-Wilk test, and equal variance tests were performed. Parametric tests were used when both normality and equal variance assumptions were met (P > 0.05). If either assumption was violated (P < 0.05), non-parametric tests were applied.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

H.-S.H. designed the experiments, wrote the manuscript, and supervised the studies. All authors reviewed, edited, and approved the manuscript. D.-F.H. carried out whole-cell patching in human samples. M.-Y.C. performed two-photon calcium imaging. Y.-H.L. conducted whole-cell patching in mouse samples. Y.-T.C. carried out ion concentration measurement. C.-K.S. provided partial grant support and technical help. K.-C.W. provided human neocortical slices, and cerebrospinal fluids from subjects with subarachnoid hemorrhage or hydrocephalus.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Mary Teena Joy and Benjamin Bessieres.

Data availability

All raw data and statistical analysis results were provided in Supplementary Data S1.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-09172-8.