Diffuse microglial responses and persistent EEG changes correlate with poor neurological outcome in a model of subarachnoid hemorrhage

Abstract

The mechanism by which subarachnoid hemorrhage (SAH) leads to chronic neurologic deficits is unclear. One possibility is that blood activates microglia to drive inflammation that leads to synaptic loss and impaired brain function. Using the endovascular perforation model of SAH in rats, we investigated short-term effects on microglia together with long-term effects on EEG and neurologic function for up to 3 months. Within the first week, microglia were increased both at the site of injury and diffusely across the cortex (2.5-fold increase in SAH compared to controls, p = 0.012). Concomitantly, EEGs from SAH animals showed focal increases in slow wave activity and diffuse reduction in fast activity. When expressed as a fast-slow spectral ratio, there were significant interactions between group and time (p < 0.001) with less ipsilateral recovery over time. EEG changes were most pronounced during the first week and correlated with neurobehavioral impairment. In vitro, the blood product hemin was sufficient to increase microglia phagocytosis nearly six-fold (p = 0.032). Immunomodulatory treatment with fingolimod after SAH reduced microglia, improved neurological function, and increased survival. These findings, which parallel many of the EEG changes seen in patients, suggest that targeting neuroinflammation could reduce long-term neurologic dysfunction following SAH.

Keywords: Electroencephalography; Fingolimod; Inflammation; Microglia; Spectral analysis; Subarachnoid hemorrhage.

Figure 1

Experimental design and MRI-based characterization of an experimental SAH model optimized for long-term study. (A) SAH was induced at day 0. MRI was performed 24 h after surgery to assess the extent and location of bleeding, followed by implantation of epidural EEG electrodes covering the convexity of the brain. Neurologic severity scoring, 24-h video EEG, and open field testing were then conducted at regular time points up to 3 months. In a separate cohort, brains were harvested at 2 and 7 days after injury to assess for degree of inflammation. Figure created with BioRender.com. (B) Kaplan–Meier survival curves are shown for sham and SAH animals for up to 7 days after surgery. Log-rank (Mantel-Cox) test, p = 0.002. Percentages represent the proportion of animals surviving by day 7. (C) Direct visualization of the ventral surface of the rat brain serves as the gold standard for confirming SAH. (D) Laser Doppler flowmetry recordings were performed over the right middle cerebral artery territory to collect regional CBF. Baseline values were obtained pre-operatively and monitored for 30 min post-operatively. A two-way ANOVA with repeated measures showed main effects of time, group, as well as an interaction between group and time (each p < 0.001). n = 14–18 per group. Data shown as median (IQR). E) SAH was identified on T2-MRI as reduced hyperintense signal at the skull base corresponding to a reduction in normal cerebrospinal fluid hyperintensity. (F–H) Ventral hypointense signal volume (V_hypo, F), total brain volume (G), and ventricular volume (H) were quantified in sham and SAH rats. n = 8–10 animals per group, independent sample t-test, data shown as mean ± standard deviation. I) ROC curve analysis confirmed the ability of V_hypo to correctly identify SAH without harvesting the brain for long-term studies. AUC = 0.963, p = 0.001. An optimal cutoff of V_hypo 4.63 mm³ corresponded to 80% sensitivity and 100% specificity for confirming SAH.

Figure 2

Early raw EEG changes and spectrogram after experimental SAH. (A) EEGs from representative sham (left) and SAH (right) animals at two days following surgery show a significant increase in slow activity throughout the brain. Data shown in referential montage, scale: 30 ms on x-axis, 200 µV on y-axis). (B) EEG spectrograms from full 24-h recordings are shown for the same animals 2 days following SAH. Compared to shams (left), SAH animals (right) exhibited an increase in overall spectral power that was most pronounced at lower frequencies, especially the delta band (1–4 Hz).

Figure 3

Widespread, persistent reduction in spectral power ratio after experimental SAH reflects changes in both fast and slow EEG frequency bands. (A) Maps of the fast-slow spectral power ratio (SPR = [α + β]/[δ + θ]) across all six electrodes are shown for sham and SAH groups over time. (B) Quantitative measurements showed a widespread reduction in SPR in the first 14 days after SAH across all electrodes. In some electrodes, reduced SPR persisted beyond 14 days and never recovered to baseline. Three-way ANOVA with main effects of time (p < 0.001), group (p < 0.001), electrode (p < 0.001) and interactions between group and time (p < 0.001). No interaction was detected between group and electrode (p = 0.062), between electrode and time (p > 0.999) or between all three variables (p > 0.999). Within each electrode, Šidák’s multiple comparisons test was used to assess for between-group differences at each time point. n = 6–10 animals per group. Data shown as mean ± standard error.

Figure 4

Increased delta power localizes to the site of hemorrhage early after experimental SAH. (A) Spectral maps of absolute delta power (1–4 Hz) across all six electrodes are shown for sham and SAH groups over time. (B) Quantitative EEG revealed increased delta power that mapped to hemorrhage areas, specifically the ipsilateral R2 and R3 and contralateral L3 electrodes which are closest to the regions overlying the bifurcation of the right internal carotid artery. Three-way ANOVA with main effects of time (p = 0.027), group (p < 0.001), electrode (p < 0.001) and interactions between group and electrode (p < 0.001). No interaction was detected between group and time (p = 0.175), electrode and time (p > 0.999), or between all three variables (p > 0.999). Within each electrode, Šidák’s multiple comparisons test was used to assess for between-group differences at each time point. n = 6–10 animals per group. Data shown as mean ± standard error.

Figure 5

SAH produces neurobehavioral deficits that correlate with changes in EEG spectral activity. (A) The 21-point modified Garcia score was performed in sham and SAH animals regularly up to 12 weeks. Deficits in SAH animals were clear early on and persisted over time (two-way ANOVA: time p < 0.001, Sham vs. SAH p < 0.001, interaction p < 0.001). (B) Representative raw data from the open field test demonstrated reduced locomotor activity in SAH animals. (C) SAH animals showed decreased ambulatory distance compared to shams. (D) No differences were seen between sham and SAH animals in the percentage of time spent in the center of the open field apparatus (n = 6–10 animals pr group. Data shown as median ± IQR). (E) Spectral power ratio (SPR = [α + β]/[δ + θ]) showed a positive correlation with neurological score (Spearman correlation: p < 0.001, r = 0.484). (F) Absolute delta power and neurological score were inversely correlated (Spearman correlation: p = 0.004 r = − 0.119. (G,H) and positive correlations were observed between (G) relative alpha variability (RAV = α / broadband, p = 0.013, r = 0.120) and (H) alpha-delta ratio (ADR = α / δ, p = 0.039, r = 0.091). Individual data points shown were overlaid with a linear regression line and 95% confidence interval.

Figure 6

SAH results in widespread microglial responses most pronounced near the site of hemorrhage. (A) Iba-1 immunohistochemistry in sham and SAH animals 2 days after surgery demonstrated increased Iba-1⁺ area in untreated SAH and SAH-FTY animals compared to shams that was more pronounced ipsilateral to the injury site at the base of the brain. In addition to increased parenchymal Iba-1⁺ staining, clots were observed in the subarachnoid space with surrounding Iba-1⁺ cells. (B) Iba-1 and Prussian Blue co-stain reveal co-localization of Iba-1⁺ cells with blood products both within the subarachnoid space (right inset, red outline) and the nearby brain parenchyma at the ipsilateral base of the brain. (C) Increased Iba-1⁺ area was also observed distant from the site of hemorrhage in the cortex, particularly on the side ipsilateral to injury in untreated SAH and SAH-FTY animals compared to sham controls.

Figure 7

Spatiotemporal mapping reveals increased microglia and cell size that are attenuated by FTY treatment. 64 regions of interest (ROIs) were selected spanning the entire rostral-caudal axis. Iba-1⁺ area per field within each ROI was then quantified and mapped onto the rodent brain in sham, untreated SAH, and SAH-FTY animals at 2- and 7-days following surgery. (A) There was a clear association between location of bleeding and Iba-1⁺ cell responses, although several distant sites also showed increased signal, typically ipsilateral to arterial perforation. At 2 days, Iba-1⁺ was increased in both untreated SAH and SAH-FTY animals compared to shams. At 7 days, SAH-FTY animals displayed a drastic reduction in Iba-1⁺ compared to untreated SAH animals. (B) Increased Iba-1⁺ was observed in the ipsilateral hemisphere of SAH animals, most pronounced on day 7, where Iba-1⁺ was significantly increased in SAH animals compared to shams and attenuated by FTY. Day 2: Two-way ANOVA with main effects of group (p = 0.032), Šídák’s multiple comparisons test for ipsilateral hemisphere: sham vs. SAH p = 0.072, SAH vs. SAH-FTY p = 0.993, for contralateral hemisphere: sham vs. SAH: p = 0.725, SAH vs. SAH-FTY p = 0.875. Day 7: Two-way ANOVA with main effects of group (p = 0.014), Šídák’s multiple comparisons test for ipsilateral hemisphere: sham vs. SAH p = 0.020, SAH vs. SAH-FTY p = 0.025, for contralateral hemisphere: sham vs. SAH: p = 0.821, SAH vs. SAH-FTY p = 0.645). (C) Similar results were observed with Iba-1⁺ cell size, where increased cell size was observed at 2 days in SAH animals compared to shams, but by day 7 SAH-FTY animals had reduced Iba-1⁺ cell size compared to untreated SAH animals. Day 2: Two-way ANOVA with main effects of group (p = 0.026) and hemisphere (p = 0.024), Šídák’s multiple comparisons test for ipsilateral hemisphere: sham vs. SAH p = 0.037, SAH vs. SAH-FTY p = 0.974, for contralateral hemisphere: sham vs. SAH: p = 0.476, SAH vs. SAH-FTY p = 0.770. Day 7: Two-way ANOVA with main effects of group (p = 0.003), Šídák’s multiple comparisons test for ipsilateral hemisphere: sham vs. SAH p = 0.330, SAH vs. SAH-FTY p = 0.013, for contralateral hemisphere: sham vs. SAH: p = 0.996, SAH vs. SAH-FTY p = 0.081. n = 4–6 animals per group. Data shown as mean ± standard error.

Figure 8

Targeting microglial activation after experimental SAH improves neurologic outcome and survival. (A–C) Primary CX₃CR-1^GFP/GFP microglia exposed to 40 µM hemin for 12 h in the presence of 2 µm fluorescent blue latex beads demonstrated increased phagocytosis in response to hemin. (A) Images of GFP-expressing microglia (green) show increased bead (blue) phagocytosis and morphological changes to a more amoeboid-shape following exposure to hemin, features consistent with microglial activation. (B,C) There was a significant increase in phagocytic activity of microglia in the presence of hemin (p < 0.001, Mann–Whitney test). (D) FTY treatment resulted in improved survival in SAH animals up to 7 days after surgery (log-rank (Mantel-Cox) test, p = 0.065). Percentages represent the proportion of animals surviving by day 7. (E) Neurologic severity score showed deficits in untreated SAH animals that were prevented with FTY treatment. Two-way ANOVA: Time: p = 0.007, Group: p < 0.001. Šídák’s multiple comparison test: sham vs. SAH p < 0.001, sham vs. SAH-FTY p = 0.541, SAH vs. SAH-FTY p < 0.001. n = 8–10 per group. Data shown as mean ± standard error. (F) Scatterplot of Iba-1 + area per field (%) versus neurological score 24 h after surgery. Each data point represents one animal. Spearman correlation test. r and p values are reported. n = 10 per group.