Neutrophils and Neutrophil Extracellular Traps Cause Vascular Occlusion and Delayed Cerebral Ischemia After Subarachnoid Hemorrhage in Mice

Abstract

Background:

After subarachnoid hemorrhage (SAH), neutrophils are deleterious and contribute to poor outcomes. Neutrophils can produce neutrophil extracellular traps (NETs) after ischemic stroke. Our hypothesis was that, after SAH, neutrophils contribute to delayed cerebral ischemia (DCI) and worse outcome via cerebrovascular occlusion by NETs.

Methods:

SAH was induced via endovascular perforation and SAH mice were given either a neutrophil depleting antibody, a PAD4 inhibitor (to prevent NETosis), DNAse-I (to degrade NETs), or a vehicle control. Mice underwent daily neurological assessment until day 7 and then euthanized for quantification of intravascular brain NETs (iNETs). Subsets of mice were used to quantify neutrophil infiltration, NETosis potential, iNETs, cerebral perfusion, and infarction. In addition, NETs markers were assessed in aneurysmal SAH patients’ blood.

Results:

In mice, SAH led to brain neutrophil infiltration within 24 hours, induced a pro-NETosis phenotype selectively in skull neutrophils, and caused a significant increase in iNETs by day 1 which persisted until at least day 7. Neutrophil depletion significantly reduced iNETs, improving cerebral perfusion, leading to less neurological deficits and less incidence of DCI (16% vs 51.9%). Similarly, PAD4 inhibition reduced iNETs, improved neurological outcome, and reduced incidence of DCI (5% vs 30%), whereas degrading NETs marginally improved outcomes. Aneurysmal SAH patients who developed DCI had elevated markers of NETs compared to non-DCI patients.

Conclusions:

After SAH, skull-derived neutrophils are primed for NETosis and there are persistent brain iNETs which correlated with delayed deficits. The findings from this study suggest that, after SAH, neutrophils and NETosis are therapeutic targets which can prevent vascular occlusion by NETs in the brain, thereby lessening the risk of DCI. Finally, NETs markers may be biomarkers which can predict which aneurysmal SAH patients are at-risk for developing DCI.

Graphical Abstract

Introduction

Aneurysmal subarachnoid hemorrhage (SAH) is a devastating disease with high morbidity and mortality.¹ In patients that survive the initial event, approximately 30% develop delayed cerebral ischemia (DCI), usually 4-10 days after SAH. DCI is the most common cause of morbidity and mortality in patients that survive the initial aneurysm rupture.² The causes of DCI are multifactorial, and include inflammation and vascular occlusion.^3,4

As an abundant immune cell type that responds early to brain injury, neutrophil involvement has been explored in many post-SAH pathological events, including early brain injury and DCI.^5–12 Recently, analysis of differentially express genes identified six neutrophil genes as signature biomarkers of SAH.¹² Experimentally, work by Provencio and others provided clear evidence that neutrophil antagonists can prevent blood-brain barrier disruption, microvascular injury, inflammation, and vasospasm, thereby improving functional and cognitive outcomes.^13–18 In fact, neutrophil depletion can improve cerebral perfusion at 3 hours post-SAH in mice.¹⁷ However, it is unknown if neutrophil antagonism can prevent DCI, and the underlying mechanisms by which neutrophils are deleterious remains unclear.

A critical function of neutrophils is the formation of neutrophil extracellular traps (NETs), large web-like scaffolds of DNA and proteins which are released to trap pathogens and fight infection. Although beneficial for pathogen removal, NETs exacerbate injury after ischemic and hemorrhagic stroke.^19–22 A key protein involved in NETosis is peptidylarginine deiminase 4 (PAD4) which prepares neutrophils for undergoing NETosis by unraveling DNA within the neutrophil.²³ As a crucial component of NETs, inhibition of PAD4 improves outcomes in experimental ischemic stroke studies.^20,22,24 In a very recent experimental paper, degrading NETs, using DNAse, was shown to improve outcomes after SAH.²⁵

Thus, we hypothesized that neutrophils contribute to DCI and worsen outcome after SAH via cerebrovascular occlusion induced by NETs, and that early prevention of neutrophil NETosis can improve outcomes. To test our hypothesis, we used a mouse model of SAH to examine 1) if depleting neutrophils can reduce NETs, improve cerebral perfusion, and prevent DCI, and 2) if preventing NETs or degrading them after SAH is therapeutically beneficial. Finally, to test for potential clinical relevance, we performed the first SAH study to determine if NETs are elevated after SAH in humans.

Materials and Methods

Animal study

Two hundred ten female C57BL/6J mice (4-6 months old) were used as SAH is slightly more predominant in women.^26,27 Animals were housed in a temperature-controlled room with a 12-hour light-dark cycle and had access to food and water. Mice were electronically randomized into groups before use. Our previous study which reports a correlation between microthrombi and DCI was used to power this study;²⁸ SigmaPlot 11.0 was used to estimate all sample sizes using data from previous experiments²⁸ with α=0.05 and β=0.2. All investigators performing functional assessment, measurement of outcomes, or data analysis were blinded to the experimental groups.

SAH was induced using the endovascular perforation model as we have done previously.²⁸ Briefly, the external, common, and internal carotid arteries were exposed. The external carotid artery was ligated, and a 5-0 monofilament suture was inserted into the stump of the external artery. The suture was advanced through the internal carotid artery and into the Circle of Willis until vessel perforation occurred. Confirmation of SAH was done through intracranial pressure monitoring^29,30 or cessation of breathing (a priori exclusion criteria).^31–33 Sham mice underwent all surgical procedures (including ligation of the external carotid artery and insertion of the monofilament) save for vessel perforation.

Neutrophil depletion:

Neutrophils were depleted using an anti-Ly6G antibody as previously described.³⁴ Briefly, 500 μg of antibody was injected intraperitoneally 1 day and 1 hour prior to SAH, and then 250 μg was given on day 3 post-SAH (n=36). SAH controls (n=36) and shams (n=14) received injections of an anti-IgG antibody.

NETs inhibition:

The PAD4 inhibitor N-α-benzoyl-N5-(2-chloro-1-iminoethyl)-l-ornithine amide (Cl-amidine) was used to inhibit NET formation. A dose of 3.33 mg/kg in normal saline was used. Cl-amidine was injected intraperitoneally either 15 minutes prior to SAH surgery (pre-treatment group, n=10) or 3 hours post-SAH (post-treatment group, n=10). Vehicle-treated mice received normal saline (n=10).

NETs degradation:

DNAse-I was used to degrade NETs. DNAse-I (in normal saline) was administered intraperitoneally 1 hour (50 μg) and 8 hours (10 μg) post-SAH (n=10).³⁵ Vehicle-treated mice received normal saline (n=10).

Neurobehavior:

Mice were assessed daily for their behavior score using a composite neuroscore which assesses sensorimotor function.³⁶ Open field and y-maze tests were performed on days 1 and 7. The open field task consisted of a single 20-minute trial in which mice freely explored a 40x40x30 cm box. Percent time spent exploring (moving) was quantified.³⁷ For the Y-maze test, mice were placed in the center of the y-maze and given 5 minutes to explore. Alteration index was calculated as (new arm visits minus old arm visits divided by total alterations).³⁸

Flow cytometry:

Neutrophil counts were determined in the brain, spleen, skull bone marrow, femur bone marrow, and blood 1, 3, 5, and 7 days after SAH. Briefly, mice (n=5/group/time-point) were deeply sedated, blood was collected, and then mice were transcardially perfused with PBS before removing the brain, skull, femur, and spleen. The whole brain, skull bone marrow, femur bone marrow, and whole spleen were processed, and flow cytometry was performed, as previously described.^37,39–41 The antibodies used were: CD45-ef450, CD11b-APC-ef780, Ly6G-FITC, Ly6C-AF700, and LIVE/DEAD aqua. Data were acquired using Cytoflex-S (Beckman Coulter) and analyzed using FlowJo. After gating for singlets and live cells, neutrophils were identified as CD45+Cd11b+Ly6G+ cells. t-distributed stochastic neighbor embedding (tSNE) plots were generated in FlowJo using DownSample plug-in (3,000 cells per sample for each study group) followed by the tSNE algorithm on all compensated parameters (except viability) at 1,000 iterations, perplexity of 30, learning rate of 5040, and Barnes-Hut gradient algorithm.

NETosis assay:

One day after SAH (n=9/group), NETosis was measured in the femur bone marrow and skull bone marrow using a previously described method.⁴² Briefly, skull and femur neutrophils were isolated and sorted (CD45+CD11b+Ly6G+ cells) using a BD FACSMelody. Sorted neutrophils were incubated with SYTOX Green (5 μL of a 1:250 dilution) and DMSO for 2 hours, then NETosing cells were identified as SYTOX+ (on a Beckman Coulter Cytoflex-S). SYTOX+ neutrophils (percent of total neutrophils) were quantified.

Immunostaining:

Brain intravascular NETs were assessed in sham (n=17), sham+IgG antibody (n=6), SAH (days 1, 2, 3, 5, and 7, n=5/time-point), SAH+IgG antibody (n=6), SAH+normal saline (n=12), SAH+Ly6G antibody (n=6), SAH+PAD4 inhibitor (pre-treatment, n=6), SAH+PAD4 inhibitor (post-treatment, n=6), and SAH+DNAse-I (n=6). PFA-fixed brains were sectioned into 40 μm thick slices. Slices at −2 from bregma were stained for NETs and blood vessels. Brain slices were imaged (DMi8, Lecia) using a 40x objective lens and individual images were processed using THUNDER, and then stitched together using the LASX software. Intravascular NETs were counted throughout the entire stitched brain slice using the LASX software. To make analysis easier, the intensity of either the blue or red channel (or both) were slightly adjusted (to make the staining appear brighter compared to background). Each blood vessel segment (red stain) which had positive NET (blue) staining was counted as an intravascular NET. Any NETs staining that occurred outside of the vessels was not counted. We assumed that if there were multiple NET (blue) segments within a single vessel (red) segment, this was a single intravascular NET. Vessel “diameter” was measured using the line tool in the LASX software during analysis.

Intravital microscopy:

The formation of cortical NETs was observed using intravital microscopy (Nikon A1R MP) in mice allocated to sham+IgG antibody (n=6), SAH+IgG antibody (n=10), and SAH+Ly6G antibody (n=10) on days 2, 4, and 6 post-SAH. Briefly, sedated mice had a stacked-glass cranial window (one 5 mm glass, three 3 mm glass)⁴³ implanted over the MCA territory (center of the window 1.7 mm right lateral from midline and 2.2 mm posterior from Bregma) at least 3 weeks prior to SAH. Following SAH, mice were imaged on days 2, 4, and 6. A cocktail containing SYTOX green (extracellular DNA marker as a surrogate for NETs, 3.6 μL) and 70 kDa red dextran (intravascular space filler, 2 mg) was intravenously injected 15 minutes prior to imaging. In a separate cohort of mice (n=2 sham, n=5 SAH+IgG antibody, and n=5 SAH+Ly6G antibody), as dextran is a flow-dependent marker, we intravenously injected a cocktail of SYTOX green and lectin (100 μL on day 2, 50 μL on days 4 and 6) 30 minutes prior to imaging. The entire cranial window was imaged (z-stack) and then z-stacked images were denoised before quantifying the area (length x width) of intravascular NETs (SYTOX+ (green) stain within/along red (vessel) stain).

MR imaging:

in-vivo MRI was performed on a 7 T system 7interfaced with ParaVision 5.1 (Bruker Biospin, Billerica, MA). Flow-Sensitive Alternating Inversion Recovery (FAIR) based sequence and diffusion tensor imaging were used to evaluate the changes of cerebral blood flow (CBF) and infarction, respectively, in mice allocated to sham+IgG antibody (n=6), SAH+IgG antibody (n=15), and SAH+Ly6G antibody (n=15). Specifically, CBF was measured using FAIR-EPI sequence with the following parameters: TR/TE: 18000/11 ms with recovery time of 10000 ms; 8 inversion times (TIs): [30 500 1000 1600 2000 3000 5000 8000] ms; resolution: 250 × 250 μm²; slice thickness: 1 mm; number of slices: 4. For diffusion tensor imaging, the following parameters were used: TR/TE: 7500 / 20 ms; 4 segments; resolution: 135 × 135 μm²; slice thickness: 1 mm; b-value: 800 s/mm² and 30 diffusion directions. Prior to SAH, mice received baseline imaging for CBF. On days 1 and 5 post-SAH, mice were subjected to MR imaging for CBF and infarction. The CBF map was calculated using in-house scripts developed in Matlab (Mathworks, MA). CBF (baseline, day 1, and day 5) and infarction volume (days 1 and 5) were analyzed for the cerebrum using ITK-SNAP.⁴⁴ The ROIs examined were the cortex, striatum, thalamus, and hippocampus for the left and right hemispheres. The Allen Brain Explorer was used to map the MRI images to the brain locations.⁴⁵ CBF data was normalized for each mouse to its baseline values for each ROI.

Histological staining for infarcts:

While we performed MRI analysis for infarction, we utilized Nissl staining of post-mortem brains (euthanized on day 7) to confirm infarction in MRI mice (n=6 sham+IgG antibody, n=9 SAH+IgG antibody, n=7 SAH+Ly6G antibody) and to assess infarction in other mice (n=3 SAH+IgG antibody, n=5 SAH+Ly6G antibody, n=12 sham, n=8 SAH+vehicle (saline administered 15 minutes before SAH), n=8 SAH+vehicle (saline administered 1 hour post-SAH), n=8 SAH+PAD4 inhibitor (pre-treatment), n=7 SAH+PAD4 inhibitor (post-treatment), and n=8 SAH+DNAse-I). Briefly, cresyl violet solution was used to stain brain slices every mm from +2 to −3 from Bregma following manufacturer recommendations. Stained slices were imaged (FLEXACAM on a DMi8, Lecia) using a 20x objective lens and individual images were then stitched together using the LASX software. Slices were then assessed for infarct area using ImageJ. Adjacent slices which had infarct at the same location (i.e. the infarct was perceived to extend from one slice to another), volume was calculated as:

$$\text{Infarct Volume}\left({\text{mm}}^3\right)\text{per Infarct for}n\text{Adjacent Slices}=\left((n+1)\text{mm}\right)\left(\frac{1}{n}\sum _{i=1}^n\text{Infarct area of lesion}\text{for slice}i\right)$$

When an infarct did not extend to adjacent slices, volume was computed as

$$\text{Infarct Volume per Infarct}=\left(1\text{mm}\right)\left(\text{Infarct area}\right)$$

These equations assume that the infarct(s) observed in a slice are at their maximum area. Total infarction volume was calculated as the sum of all k infarct volumes throughout the slices measured.

Delayed neurological decline:

Delayed neurological decline (DND) is one component of DCI pathology.⁴⁶ Specifically, DND is characterized as a delayed decline in neurological behavior and is observed in humans⁴⁶ and mice.²⁸ In a previous paper, we characterized DND in mice as follows: 1) mice must have some recovery of behavioral performance (using neuroscore in our study) from the neuroscore on day 1; 2) after some recovery, mice experiencing a neuroscore which is 5 or more points less than their best performance (from any prior post-SAH day).²⁸ Animals experiencing delayed death (after some functional recovery) were also considered as developing DND since their neuroscore would be equal to 0. The neutrophil depletion study was powered for DND analysis. To increase power and sample size for the PAD4 inhibitor and DNAse-I studies, we combined the vehicle-treated SAH mice (normal saline 15 minutes before SAH and normal saline given 1 and 8 hours post-SAH, n=10/group) and combined the two PAD4 inhibitor groups (n=10/group). DND incidence was analyzed by a log-rank test. Table S1 shows representative data for mice developing DND and those not developing DND.

$$\text{DND Incidence}(\%)=\frac{\left(\%\text{of animals that experienced delayed behavioral decline in the neuroscore test of 5 points or more}\right)}{(\text{total number of animals surviving more than}2\text{days})}x100$$

Statistical analysis:

Data is presented as individual data points with the mean and standard deviation (unless otherwise stated). All data was assessed for normality and homoscedasticity (prior to analysis) using the Shapiro-Wilk test and the Levene statistic. A p<0.05 was considered statistically significant. One-way or two-way ANOVAs (with repeated measures as appropriate) and two-tailed unpaired t-tests were performed. Correlational analysis for NETs counts, infarction (DWI), infarction (Nissl), CBF, and DND were performed using Spearman and Pearson correlational coefficients as appropriate. Graphpad Prism and SPSS 28 were used for graphing and analysis. Mortality and excluded animals are reported in the Supplemental Material. Statistical reports for all outcomes can be found in the Tables S7-S37. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Human study

Consecutive patients experiencing SAH who were seen at the Memorial Hermann Hospital in the Texas Medical Center between December 2018 and August 2021 were enrolled in a prospective observational study. After excluding patients not having SAH from an aneurysm rupture (confirmed via imaging) and COVID+ patients, one hundred twenty-seven patients were included in this study. Modified Fisher and Hunt Hess were documented. Patients were treated by either aneurysm clipping or coiling, and all patients received nimodipine and corticosteroids as standard of care. Patients given anti-platelets were documented. Human blood samples (n=127 SAH patients) were collected on days 1, 2, 4, 7, and 10 after SAH by the NeuroICU staff. Blood was collected into sterile vacutubes (EDTA-coated), processed within one hour (centrifuged to obtain plasma), deidentified, and stored in a secure, authorized access only, −80°C freezer. DCI was determined prospectively (adjudicated by three clinicians in weekly meetings) as a neurological decline (Glasgow Coma Scale) of 2 or more points after exclusion of confounding variables (e.g. infection, hypotension, seizures). Healthy controls (n=6, powered a priori) had blood collected one time.

Plasma samples were thawed immediately before use. Plasma samples were diluted and processed in a blinded fashion following manufacture protocols to assess two markers of NETs using ELISA: neutrophil elastase and citrullinated histone H3.

For a preliminary assessment of NETs in human brains, two brains were collected from aneurysmal SAH patients via autopsy (one on day 1 and one on day 12). Tissue was collected from 12 regions in the brain,⁴⁷ embedded in paraffin, and then sectioned into 4 μm thick slices. Slices were deparaffinized, then underwent antigen retrieval before being stained overnight in primary antibodies (citrullinated histone H2B, laminin). Secondary antibodies (1:200) were applied for 1 hour at room temperature and then slices were mounted.

Statistical analysis:

Individuals performing statistical analysis were blinded to outcome and patient demographics. Descriptive statistics such as mean (standard deviation) and frequency (percentage) were used for continuous and categorical variables, respectively. T-tests and Fisher’s exact tests were used for testing the differences of the variable distributions between the DCI and non-DCI groups. Multivariable logistic regression models were used to predict the effect of the NETs markers on DCI. In the multivariable regression model, all demographics with a p<0.1 between no DCI and DCI patients were treated as possible confounding factors and were included in the models. Statistically significant difference was defined as a p-value <0.05. ROC curves were generated using Graphpad Prism. All analyses were conducted using R version 4.2.1.

Study approval

Animal experiments were approved by the UTHealth Animal Welfare Committee, conducted in compliance with the National Institutes of Health Guidelines for the Use of Animals in Neuroscience Research, and are reported in compliance with the ARRIVE guidelines. Aneurysmal SAH Patients were enrolled (via written informed consent prior to participation) in the study under a protocol approved by the UTHealth Institutional Review Board.

Results

Neutrophils infiltrate the brain within 24 hours after SAH

To gain a generalized understanding of neutrophil behavior after SAH, we performed flow cytometry of blood, spleen, peripheral bone marrow, skull, and brain (Figure 1A-C) to determine if SAH leads to changes to neutrophil counts. There was significantly higher neutrophil infiltration in the brain on day 1 compared to sham which returned to baseline by day 5 (Figure 1B). Similarly, neutrophil counts were elevated in the blood and spleen on day 1 which normalized by day 3 (Figure 1C). While SAH did not cause an increase in neutrophil counts of either the peripheral or skull bone marrow on day 1, there were delayed increased neutrophil counts (Figures 1C and S3, Table S2). A tSNE plot shows the relative changes in live CD45+ cells in the brain over time after SAH (vs sham) (Figure 1D).

Figure 1. Flow Cytometry Analysis of Neutrophil Counts after SAH in Mice.

A. Gating strategy to define neutrophil population (following identification of singlets and live cells by FSC vs SSC and LIVE/DEAD aqua, respectively). Percentages of neutrophils was defined as the number of neutrophils (CD45^highCD11b+Ly6G+ cells) divided by the total number of live CD45^highCD11b+ cells. Actual cell counts are reported in Table S2. B. SAH induces neutrophil infiltration into the brain by day 1. This response normalizes by day 5. n=5/group/time-point. One-way ANOVA with Tukey post hoc. C. Longitudinal neutrophil counts (CD45^highCD11b+Ly6G+) in the blood, spleen, peripheral bone marrow, and skull. n=5/group/time-point. One-way ANOVA with Tukey post hoc: blood, peripheral bone marrow, and skull. Kruskal-Wallis with Dunn’s post hoc: spleen. D. tSNE plots for visual representation of live CD45+ cell counts over time.

iNETs form within 1 day after SAH and persist until at least day 7

As NETosis is a critical function of neutrophils, and since NETs have been reported to cause cerebrovascular occlusion after ischemic stroke,^48,49 we quantified brain iNETs following SAH on days 1, 3, 5, and 7. Using quantitative immunohistochemical methods, we found significantly higher levels of intravascular NETs in the brain tissue 1 and 7 days post-SAH compared to sham (Figures 2A-B, S5-S8). Interestingly, NETs tended to cluster together in the brains following SAH in mice, leaving some areas of the brain nearly free of NETs (Figures S9-S11). Within those clusters, iNETs affected about 83% of blood vessels with diameters less than 20 μm. In contrast, iNETs were observed in less than 1% of vessels larger than 20 μm throughout the brain slice. Finally, the clusters of iNETs did not localize to specific regions of the brain after SAH, but the area affected by iNETs clusters positively correlated with total iNETs count (p<0.001) (Tables S3 and S5).

Figure 2. SAH Increases Brain Intravascular NETs and Skull Neutrophil NETosis in Mice.

A. Representative images from mice brains on day 7 post-SAH stained for citrullinated histone (NET marker, blue) and blood vessels (laminin, red). The right image is a zoomed portion of the SAH Day 7 image (white box) showing that vessels less than 20 μm are occluded by NETs. Scale bar = 50 μm. Images are taken from the thalamus region. B. Quantification of intravascular brain immunostaining for NETs (co-localization of blue in red). n=5/group. Kruskal-Wallis with Dunn’s post hoc. C. Following SAH, skull neutrophils have higher NETosis than femur neutrophils. NETosis was assessed using flow cytometry analysis of sorted neutrophils from either skull bone marrow or femur bone marrow. Sorted neutrophils were subjected to 2 hours of culturing before assessing for NETosis. n=8-9/group. Unpaired two-tailed t-test. BM: bone marrow.

Skull-derived neutrophils have an increased NETosis response after SAH, but peripheral neutrophils do not

Since SAH caused a sustained increase in iNETs, we examined the NETosis potential of neutrophils after SAH using neutrophils from the bone marrow. As, the skull was recently shown to be a unique reservoir for neutrophils that preferentially responded to stroke injury,⁵⁰ we tested the NETosis potential of neutrophils collected from the peripheral bone marrow (femur) and the skull bone marrow. To measure the NETosis potential of neutrophils following SAH, we employed an in vitro assay using neutrophil stimulation and DNA labeling (SYTOX green).⁴² The skull and femur bone marrow neutrophils were isolated from sham and SAH mice on 1 day. SAH caused skull neutrophils to have a significantly higher NETosis potential compared to that of sham (Figure S4). Unexpectedly, neutrophils from the femur bone marrow of SAH mice showed the opposite – reduced NETosis propensity (Figure 2C).

Neutrophil depletion reduces iNETs and improves outcome after SAH

As neutrophils infiltrate the brain and there is a significant number of brain vessels occluded by iNETs, we sought to determine if neutrophil depletion prior to SAH can prevent iNETs. SAH in mice with Ly6G-Ab-mediated depleted neutrophils resulted in fewer brain iNETs as compared to non-neutrophil depleted animals (Figure 3A) and this reduction in iNETs formation coincided with significantly reduced neurological dysfunction (Figure 3B). Specifically, neutrophil depletion (compared to non-neutrophil depleted SAH mice) improved sensorimotor function (neuroscore), day 1 working memory (Y-maze), and improved exploration (open field) on day 7.

Figure 3. Neutrophil Depletion Prevents iNETs and Improves Neurological Behavior after SAH in Mice.

A. Representative images from mice brains on day 7 post-SAH stained for citrullinated histone (NET marker, blue) and blood vessels (laminin, red). Scale bar = 50 μm. Quantification of immunostaining for iNETs (co-localized blue and red staining) 1 and 7 days post-SAH. Day 1: n=6/group. Day 7: n=6 sham, n=12 for each SAH group. One-way ANOVA with Tukey post hoc. Images are taken from the cerebral cortex region. B. Neutrophil depletion improves behavioral performance after SAH assessed using a composite neuroscore, Y-maze, and open field. D1 neuroscore: Kruskal-Wallis with Bonferroni post hoc. D1-7 neuroscore: Friedman test followed by Kruskal-Wallis with Dunn’s post hoc for each time-point. Y-maze: One-way ANOVA with LSD post hoc. Open field Day 1: One-way ANOVA with Tukey post hoc; open field Day 7: Kruskal-Wallis with Dunn’s post hoc. * Sham vs SAH+IgG Antibody Day 1 p<0.001, Day 2 p<0.001, Day 3 p<0.001, Day 4 p=0.007, Day 5 p=0.018, Day 6 p=0.017, Day 7 p=0.008. † Sham vs SAH+Ly6G Antibody Day 3 p=0.020. # SAH+IgG Antibody vs SAH+Ly6G Antibody Day 1 p<0.001, Day 2 p=0.045, Day 7 p=0.005. Individual mice performance on the 7-day neuroscore are plotted in Figure S13A.

As DCI is a severe clinical consequence that affects about one third of aneurysmal SAH patients, we performed MRI at baseline and at days 1 and 5 post-SAH to determine if neutrophil depletion could restore CBF and reduce infarct development (markers of DCI). Compared to non-neutrophil depleted mice, neutrophil depletion significantly improved cerebral perfusion 5 days post-SAH (Table S4) and significantly reduced delayed infarct volume (assess using DWI) (Figures 4A-C, S16-S17). Since animals were euthanized on day 7, we also performed Nissl staining to assess for infarction. Infarct quantification on day 7 post-SAH revealed that neutrophil depleted SAH mice had less infarction than non-depleted SAH mice (Figure 4D-E). Interestingly, iNETs count inversely correlated with CBF; higher iNETs count was associated with lower CBF of the cortex (p=0.049) and striatum (p=0.014) (Table S5). There was also a correlation between iNETs count and day 7 infarct volume (p=0.004) (Table S5). Finally, clusters of iNETs were observed in infarcts (Figure S12).

Figure 4. Neutrophil Depletion Leads to Improved CBF and Prevents Delayed Cerebral Infarction.

A. Representative images of CBF before SAH (Baseline), 1 day, and 5 days post-SAH. B. Quantification of CBF of the striatum from FAIR MRI. Mean and SEM are plotted. n=6 sham, n=12-13 SAH+IgG Antibody, n=11-12 SAH+Ly6G Antibody. Repeated measures two-way ANOVA analysis with LSD post hoc. At indicated time-point: * Sham vs SAH+IgG Antibody p=0.0, # Sham vs SAH+Ly6G Antibody p=0.0, † SAH+IgG Antibody vs SAH+Ly6G Antibody p=0.0. Plots of the changes for individual mice are plotted in Figure S17. C. Quantification of infarction 1 and 5 days post-SAH from DWI MRI. Mean and SEM are plotted. n=6 sham, n=13-14 SAH+IgG Antibody, n=11-12 SAH+Ly6G Antibody. Kruskal-Wallis with Dunn’s post hoc for each time-point. At indicated time-point: * p=0.012 Sham vs SAH+IgG Antibody and p=0.045 Sham vs SAH+Ly6G Antibody, # p=0.019 Sham vs SAH+IgG Antibody, † p=0.048 SAH+IgG Antibody vs SAH+Ly6G Antibody. Data for individual mice are plotted in Figure S16. D. Representative images of Nissl-stained brains 7 days post-SAH. E. Neutrophil depletion reduces infarction on day 7 (assessed in Nissl-stained slices). n=6 sham, n=11 for SAH groups. One-way ANOVA with Tukey post hoc. Mean and SD are plotted. All sham mice received IgG antibody injections.

Since neutrophil depletion improved cerebral perfusion and reduced infarction, we tested if depletion was related to reduced formation of NETs using in-vivo intravital imaging of cortical vasculature on days 2, 4, and 6 following SAH (Figures 5A, S14-S15). Quantitative assessment of the images revealed increased iNETs area on day 6 in non-depleted SAH mice, whereas the area of iNETs was reduced in neutrophil depleted SAH mice (Figure 5B). iNETs were observed to occlude vessels with diameters less than 15 μm. In vessels with diameters 15-30 μm, iNETs were observed adhering to the vessel wall but not occluding flow.

Figure 5. Neutrophil Depletion Reduces in-vivo iNETs Formation and Incidence of Delayed Neurological Decline (DND).

A. Representative images of iNETs taken during in-vivo intravital microscopy of the cortical MCA territory (center of the window 1.7 mm right lateral from midline and 2.2 mm posterior from Bregma). NETs (green/yellow from SYTOX staining) are visible in the vessels (red from intravascular red dextran) of SAH mice but not sham mice. SYTOX green was used to label DNA as a marker of NETs. Scale bar = 100 μm. Images shown EDF focused from the entire denoised z-stack for a representative area within the cranial window. B. Quantification of iNETs area from intravital imaging. Individual values are plotted. n=8 sham, n=11-15 SAH+IgG Antibody, n=12-14 SAH+Ly6G Antibody. Repeated measures two-way ANOVA on ranks (Friedman test) following by Kruskal-Wallis with Dunn’s post hoc for each time-point. At indicated time-point: * Sham vs SAH+IgG Antibody Day 2 p=0.022, Day 4 p=0.048, Day 6 p=0.005. † Sham vs SAH+Ly6G Antibody Day 2 p=0.038 and Day 4 p=0.037. C. Kaplan-Meier plot for development of DND (% of animals that experienced delayed behavioral decline in the neuroscore test). n=25-27/group. Mantel-Cox test. p=0.0103. All sham mice received IgG antibody injections.

Finally, we evaluated Ly6G antibody-mediated neutrophil depletion for the ability to reduce incidence of DND, another hallmark of DCI.⁴⁶ Neutrophil depletion led to a significant reduction in DND incidence compared to that of IgG-treated (non-depleted) SAH mice (p=0.011, Figure 4C). Interestingly, DND incidence positively correlated with iNETs count (p=0.002) and negatively correlated with CBF (cortex: p=0.052, striatum: p=0.048, thalamus: p=0.048, hippocampus: p=0.158). DND positively correlated with infarct volume on day 7 (Nissl, p=0.007) (Table S5). A tendency was observed for DND to positively correlate with infarct on day 5 (DWI, p=0.253).

Inhibition of NETosis improves outcomes after SAH, while degradation of NETs had marginal benefit

Next, as PAD4 has been suggested as a major component causing NETosis, we determined if preventing NETosis using a PAD4 inhibitor could reduce iNETs and improve neurological outcome after SAH. The PAD4 inhibitor, administered either before SAH (pre-treated) or 3 hours after (post-treated) SAH, significantly attenuated the formation of brain iNETs (Figure 6A) and led to significant improvement in sensorimotor function (neuroscore), and improved working memory (Y-maze) and exploration (open field) on day 7 (Figure 6B-D). Similarly, PAD4 inhibition led to significantly reduced brain infarct volume 7 days post-SAH (Figure 6E-F). More importantly, PAD4 inhibition significantly reduced the incidence of DND compared to vehicle-treated SAH mice (p=0.0389, Figure 6G).

Figure 6. PAD4 Inhibition Prevents iNETs and Improves Outcome after SAH in Mice.

A. Representative images from mice brains 7 days post-SAH stained for citrullinated histone (NET marker, blue) and blood vessels (laminin, red). Scale bar = 50 μm. Quantification of immunostaining for iNETs (co-localized blue and red staining) 7 days post-SAH. n=6-8/group. Kruskal-Wallis with Dunn’s post hoc. Images are taken from the cortex region. B-D. Both pre-treatment and post-treatment with a PAD4 inhibitor significantly improves behavioral performance after SAH assessed using a composite neuroscore (B), Y-maze (C), and open field (D). The PAD4 inhibitor post-treatment group was not assessed using the Y-maze or open field tasks. n=8-10/group. D1 Neuroscore: Kruskal-Wallis with Bonferroni post hoc. D1-7 Neuroscore: Friedman test followed by Kruskal-Wallis with Dunn’s post hoc for each time-point. Y-maze: One-way ANOVA with LSD post hoc. Open field Day 1: One-way ANOVA with LSD post hoc. Open field Day 7: Kruskal-Wallis with Dunn’s post hoc. * Sham vs SAH+Vehicle Day 1 p<0.001, Day 2 p<0.001, Day 3 p=0.043, Day 5 p=0.002, Day 6 p<0.001. † Sham vs SAH+PAD4 Inhibitor (Pre-Treat) p=0.043, # SAH+Vehicle vs SAH+PAD4 Inhibitor (Pre-Treat) Day 1 p<0.001, Day 2 p<0.001, Day 6 p=0.045. ‡ SAH+Vehicle vs SAH+PAD4 Inhibitor (Post-Treat) Day 1 p=0.014, Day 2 p=0.007, Day 5 p=0.008, Day 6 p<0.001. E. Representative images of Nissl-stained brains 7 days post-SAH to assess for infarcts (outlined in yellow). F. PAD4 inhibition prevents the development of infarcts 7 days after SAH (assessed in the Nissl-stained brain slices). n=6-8/group. One-way ANOVA with Tukey post hoc Individual mice performance on the 7-day neuroscore are plotted in Figure S13B. G. Kaplan-Meier plot of DND incidence (% of animals that experienced delayed behavioral decline in the neuroscore test). The SAH+Vehicle group (n=20/group) combines data from the SAH+Vehicle (saline, 15 minutes before SAH) (n=10) and SAH+Vehicle (saline, 1 and 8 hours post-SAH) (n=10) groups. The PAD4 Inhibitor group (n=20/group) combines data from the SAH+PAD4 Inhibitor (Pre-Treat, 15 minutes before SAH) (n=10) and SAH+PAD4 Inhibitor (Post-Treat, 3 hours post-SAH) (n=10) groups. n=10 for SAH+DNAse-I. Mantel-Cox test. SAH+Vehicle vs SAH+PAD4 Inhibitor: p=0.0389. SAH+Vehicle vs DNAse-I: p=0.9009. All plots are SD and mean except Kaplan-Meier curve.

As inhibiting NETs formation using a PAD4 antagonist was therapeutically beneficial, we then wanted to investigate if degrading existing iNETs could also provide a benefit. To test this, we treated animals after SAH with DNAse-I as an approach to degrade strings of DNA which are the structural foundation of NETs. As anticipated, SAH increased day 7 brain intravascular NETs counts and DNAse-I reduced iNETs to a level statistically indistinguishable from sham (Figure 7A). DNAse-I treatment had a marginal benefit on neurological performance (Figure 7B-D). On day 1, DNAse-I improved sensorimotor function (neuroscore) such that it was not statistically different than that of sham mice. DNAse-I also reversed the working memory (Y-maze) dysfunction caused by SAH on day 7. While DNAse-I partially reduced iNETs counts and marginally improved neurological function, DNAse-I treatment did not reduce infarct volume (Figure 7E) nor DND incidence (p=0.901, Figure 6G).

Figure 7. DNAse-I Treatment Reduces iNETs and Marginally Improves Neurological Outcome after SAH in Mice.

A. Representative images from mice brains 7 days post-SAH stained for citrullinated histone (NET marker, blue) and blood vessels (laminin, red). Scale bar = 50 μm. DNAse-I partially prevents iNETs (quantified as co-localization of blue and red staining) 7 days post-SAH. n=6-8/group. Kruskal-Wallis with Dunn’s post hoc. Images are from the thalamus region. B-D. DNAse-I treatment marginally improves neurological performance after SAH assessed using a composite neuroscore (B), Y-maze (C), and open field (D). n=8 sham, n=7-10 SAH+Vehicle, n=5-10 SAH+DNAse-I. D1 neuroscore: Kruskal-Wallis with Bonferroni post hoc. D1-7 neuroscore: Friedman test followed by Kruskal-Wallis with Dunn’s post hoc for each time-point. Y-maze: One-way ANOVA with LSD post hoc. Open field Day 1: Kruskal-Wallis with Dunn’s post hoc. * Sham vs SAH+Vehicle Day 1 p<0.001, Day 2 p=0.033, Day 3 p=0.019, Day 4 p=0.015. † Sham vs SAH+DNAse-I p=0.044. Individual mice performance on the 7-day neuroscore are plotted in Figure S13C. E. Representative Nissl-stained brains and quantified infarct volume on day 7 post-SAH. n=6-8/group. Kruskal-Wallis with Dunn’s post hoc.

Patients that developed DCI have significantly higher plasma markers of NETs

Encouraged by our animal results which suggests iNETs as pathogenic and a therapeutic target to improve both cerebral patency and neurological outcome, we determined if SAH leads to the development of NETs in humans. One hundred twenty-seven SAH patients were enrolled in the study. Of these, 40 patients developed DCI while 87 did not. Significantly more DCI patients were treated with anti-platelets during their hospital stay (p=0.035) (Table S6). Antiplatelet treatment was not specifically administered as an intervention for DCI, but rather given as a prophylactic for deep vein thrombosis in some patients. There was a trend for increased DCI in females (p=0.090) and in patients that were treated with endovascular coiling (p=0.052 vs clipping).

Herein, we examined the levels of neutrophil elastase and citrullinated histone H3 as surrogate markers for NETs. In patients, SAH caused a massive elevation in plasma levels of neutrophil elastase and citrullinated histone H3 compared to healthy controls. This suggests that the neutrophil response and NETs formation is rapid and occurs within hours after SAH onset, but also that this process continues for at least 10 days. More interestingly, after including possible confounders (sex, clipped vs coiled, anti-platelet during stay) in the multivariate regression models, patients who developed DCI had significantly higher levels of plasma neutrophil elastase on days 2, 7, and 10 post-SAH compared to those who did not develop DCI (Day 2 p=0.033, Day 4 p=0.066, Day 7 p=0.003, Day 10 p<0.001) (Figure 8A). In patients developing DCI, the concentration of neutrophil elastase continued to increase from day 1 values over the duration of the study. Citrullinated histone H3 was not significantly different between patients with DCI and those who did not develop DCI until day 10 (p<0.001) (Figure 8B). When normalized to day 1 values, the plasma levels of neutrophil elastase were significantly higher on the timepoint immediately preceding DCI onset in patients experiencing DCI than those who did not develop DCI (p=0.021) (Figure 8C). As SAH severity may have an effect on the levels of neutrophil elastase or citrullinated histone H3, we performed a subgroup analysis of these two NETs markers after stratifying the patients into either arrival Hunt-Hess grade 1-3 or Hunt-Hess 4-5. Overall, Hunt-Hess on arrival did not significantly affect the plasma levels of either neutrophil elastase or citrullinated histone H3 for non-DCI patients or DCI patients (Figure S26, Tables S34 and S35).

Figure 8. Markers of NETs are Elevated in the Plasma of Aneurysmal SAH Patients Developing DCI.

A. Patients with DCI (n=40) had higher plasma levels of neutrophil elastase on days 2, 7, and 10 compared to patients not developing DCI (n=87). At indicated time, * p=0.033, # p=0.003, † p<0.01. B. Patients with DCI had higher plasma levels of citrullinated histone H3 on day 10 compared to patients not developing DCI. Univariate analysis with repeated measures for neutrophil elastase and citrullinated histone H3 longitudinal data. C. Neutrophil elastase plasma concentration at the time-point preceding DCI onset (normalized to the day 1 value) indicates that DCI patients (n=21) have a significantly higher ratio than non-DCI patients (n=42). Unpaired two-tailed t-test. Individual data points are plotted for all three figures in Figure S24.

Finally, we assessed brain tissue collected from two aneurysmal SAH patients for intravascular NETs. One sample is from day 1 post-rupture (male, 61 years old, Hunt-Hess of 5, GCS of 3, comfort care) who did not develop DCI. The other sample is from day 12 post-rupture (female, 68 years old, Hunt-Hess of 3, GCS of 14, comfort care) who developed DCI. The tissue from the patient who developed DCI displayed intravascular NETs which occluded the vessels (Figure 9A). However, no NETs were observed in the SAH patient who did not develop DCI (Figure 9B).

Figure 9. Aneurysmal SAH Causes Brain Intravascular NETs in a Patient Developing DCI.

Representative images from the brains of a DCI patient (A, post-rupture day 12) and non-DCI patient (B, post-rupture day 1). Tissue displays NETs (white arrowheads) (blue, citrullinated histone) inside blood vessels (red, laminin) for the patient with DCI. However, no iNETs were observed in the patient who did not develop DCI. Vessels not containing NETs are indicated by a yellow arrowhead. Scale bar = 100 μm.

Discussion

Our study systematically examined the role of neutrophils and intravascular brain NETs following SAH. To date, there remains no effective therapy for improving outcomes following aneurysmal SAH for patients.¹ Inflammation is a significant contributor to DCI in SAH patients^52–54 and neutrophils are a critical immune cell type in DCI development.⁵ Furthermore, neutrophil extracellular traps play a detrimental role in other cerebrovascular pathologies, including ischemic and hemorrhagic stroke.^{19, 24} Our study tested the hypothesis that neutrophils contribute to DCI and worsen outcome after SAH via vascular occlusion by NETs. We also assessed if neutrophil depletion, NETosis inhibition, or NETs degradation could improve outcomes and reduce DCI (defined as delayed cerebral ischemia, delayed neurological decline, or delayed death).

The role of NETs has not yet been studied after SAH in humans. Herein, neutrophil elastase and citrullinated histone H3 were chosen as markers of NETs. Upon activation, neutrophil degranulation releases neutrophil elastase which is a major component of NETs.⁵¹ Citrullinated histones are also primary components of NETs. In reporting the results of the first prospective study on NETs in humans with SAH, we observed that markers of NETs are elevated in patients who develop DCI (compared to the patients not developing DCI). Specifically, neutrophil elastase is significantly elevated on days 2, 7, and 10 in post-SAH patients who developed DCI. Further analysis of this data using ROC curves suggests that neutrophil elastase may be a potential biomarker for patients who are at-risk for developing DCI (Figure S25, Table S33). On the other hand, citrullinated histone H3 was elevated in DCI patients only on day 10 which is after DCI onset in some patients. Thus, citrullinated histone H3 may not be a useful predictive biomarker to identify which SAH patients are at-risk for developing DCI (Figure S25, Table S33). This finding also suggests that NETs have formed by day 10 and may have already led to pathological damage. Our finding that arrival Hunt-Hess score did not impact the levels of either neutrophil elastase or citrullinated histone H3 suggests that these markers may be independent of the initial SAH severity which increases their potential utility as biomarkers for identifying which SAH patient is at-risk for developing DCI.

Recent evidence has shown that there are two reservoirs of neutrophils that respond to brain injury (in addition to blood neutrophils): systemic bone marrow and skull bone marrow. Interestingly, there are channels for skull neutrophils to directly access the brain.^50,55,56 In fact, Herisson et al. observed that skull neutrophils rapidly infiltrate the brain after ischemic stroke, and more neutrophils are released from the skull than the peripheral bone marrow after ischemic stroke.⁵⁰ In light of this, we determined if skull neutrophils are involved in formation of intravascular brain NETs after SAH. For the first time, the NETosis potential of neutrophils from the skull was assessed and compared to that of the peripheral bone marrow. We found that neutrophils from the skull bone marrow led to NETosis more readily than femur neutrophils after SAH. Furthermore, skull neutrophils from both sham and SAH mice have greater NETosis potential than femur neutrophils from sham or SAH mice. In fact, sham skull neutrophils have greater NETosis potential than SAH femur neutrophils (p<0.001, Figure S4). Intriguingly, the data also indicates that femur neutrophils from SAH mice have significantly reduced NETosis potential compared to that of neutrophils from sham femur neutrophils. Future work needs to be performed to determine if skull neutrophils are the primary source of iNETs in the brain. Additionally, we observed a large infiltration of viable neutrophils into the murine brain within the first 24 hours after SAH which resolves by day 5. This is strikingly different than other cerebrovascular pathologies which demonstrate a neutrophil peak around day 3.^19–21 As such, early targeting of neutrophils may be important to prevent their deleterious effects after SAH. However, this also needs to be studied in future experiments. Moreover, this raises the question as to why neutrophils reduce over time in the brain: do the neutrophils exit from the brain or do they die, potentially via NETosis?

The third novel finding from this work investigates the pathological role of NETs after SAH which, until very recently, was not examined. Hao et al. observed that citrullinated histone H3 (as a marker of NETs) was significantly elevated in the brain within 4 hours after SAH and seemed to peak at 24 hours although elevation persisted at 2 days (which was the last time-point studied) compared to sham.²⁵ The authors observed that neutrophil depletion improved day 1 outcomes (reduced brain citrullinated histone H3, lessened neuronal degeneration, reduced brain water content, improved blood-brain barrier integrity, and improved sensorimotor function). While citrullinated histone H3 was quantified using Western blot, the number of intravascular brain NETs was not measured. Our findings support the study by Hao et al. but our 7-day experiments add that neutrophil depletion can 1) reduce accumulation of intravascular NETs in the brain, 2) improve day 5 cerebral perfusion, 3) improve neurological behavior including working memory and exploration, and 4) reduces the incidence of DCI. In their series of landmark papers on the deleterious roles of neutrophils after SAH, Provencio et al. reported that neutrophil depletion can lessen delayed functional deficits on days 7-10 post-SAH using the Barnes maze task^15,16 which supports our findings. Yet, additional experiments are needed to validate our findings.

Herein we observed not only increased numbers of iNETs after SAH which correlated with DND (p=0.002), but also that increased numbers of iNETs led to iNETs clusters (p<0.001). A positive correlation was also observed between iNETs clusters and DND, suggesting that iNETs clusters are more pathological than non-clustered iNETs. Furthermore, we also observed strong evidence that the iNETs are pathological contributors to DCI as suggested by correlations between iNETs and CBF, infarction, and DND. More specifically, iNETs localization correlated with reduced CBF in those brain regions; for example, mice with significant iNETs burden of the striatum had lower striatum CBF. Cortical iNETs were observed in SAH mice using intravital microscopy and immunostaining which correlated with lower CBF in the cortex of these mice.

An important finding to mention is that, similar to microthrombi counts after SAH,²⁸ we observed that iNETs counts had high variability for the SAH, SAH+IgG Antibody, and SAH+Vehicle groups. This contrasts with low variability in iNETs counts for the treatment groups (Ly6G antibody, PAD4 inhibitor, and DNAse-I) which is likely due to the treatments reducing iNETs. As for the SAH, SAH+IgG Antibody, and SAH+Vehicle groups, the high variability is likely caused by the fact that some mice developed DND/DCI and other mice did not. This is supported as iNETs counts positively correlated with DND incidence and infarct. Furthermore, there was no correlation between iNETs counts and day 1 neuroscore or ICP (Table S5), suggesting that SAH and early brain injury are not causing the high variability in iNETs counts.

Since 1) iNETs were elevated in mice after SAH, 2) markers of NETs are predictive of patients at-risk for DCI, and 3) skull neutrophils are primed for NETosis, we hypothesized that preventing iNETs would be a viable therapeutic target. As neutrophil depletion may not be a clinically relevant approach due to infectious concerns, this is the first study to examine if preventing NETosis is a treatment for SAH. We tested two treatment regimens using a PAD4 inhibitor. Pre-treatment with a PAD4 inhibitor significantly prevented iNETs 7 days after SAH, which corresponded to improved sensorimotor and working memory function. Remarkably, treating 3 hours post-SAH had similar effects and these treatment benefits lasted for 7 days. Moreover, analysis of DND incidence indicated that treatment with a PAD4 inhibitor could significantly reduce the risk of DND (delayed deficits and delayed death) for SAH mice. Interestingly, a recent human gene analysis study reported that the PAD4 gene is highly expressed after SAH suggesting a possible role in the pathophysiological events following SAH.¹²

We then investigated if degrading iNETs (using DNAse-I) is a potential treatment for SAH. DNAse-I reduced iNETs and provided marginal improvement of neurological performance. This is in contrast to the findings of Hao et al. who observed that DNAse-I provided a therapeutic benefit, reducing blood-brain barrier disruption and neuronal degeneration, thereby improving functional outcome 24 hours post-SAH.²⁵ However, the study by Hao et al. assigned 24 mice/group whereas we had 10 mice per SAH group, and power analysis of our data suggests that DNAse-I would significantly lower iNETs counts compared to SAH+Vehicle if we added a few more mice. Thus, DNAse-I may have therapeutic potential. However, the dosing regimen used in this study did not lower the DND incidence (p=0.901 vs vehicle-treated mice). As such, administering DNAse-I daily, or a different dose, may have some therapeutic benefit, but this remains to be tested. Additionally, while DNAse-I may have a therapeutic benefit, we observed that inhibiting NETosis with a PAD4 antagonist had a much stronger treatment benefit than DNAse-I, suggesting that PAD4 inhibition is a better treatment (Table S37). Furthermore, the PAD4 antagonist was administered 3 hours post-SAH whereas DNAse-I was administered earlier (starting 1-hour post-SAH), which may make it a more promising therapeutic target.

Pathophysiological events caused by NETs

NETs have been shown to be key players in causing/promoting systemic inflammation,^57,58 neuroinflammation and microglia activation after CNS injury,^59,60 vascular instability of intracranial aneurysms,⁶¹ blood-brain barrier dysfunction after CNS injury,^20,62 and coagulation.^25,63–65 In relation to DCI after SAH, inflammation, neuroinflammation, and coagulation are all known factors contributing to DCI and poor outcome.^46,54,66,67 Although these pathophysiological events were not the focus of this study, preventing NETs after SAH via PAD4 inhibition or neutrophil depletion may attenuate or prevent these deleterious events, thereby lessening their contributions to DCI. Future studies are needed to specifically investigate the role of NETs as a cause of these other pathophysiological events after SAH.

Clinical implications

Studies by independent groups indicate that neutrophils or NETs are a therapeutic target to improve SAH outcome in experimental models.^15,16,25 Clinical findings that neutrophil levels correlate with outcome,^5–10 as well as the findings in our prospective study of 127 aneurysmal SAH patients that markers of NETs are significantly elevated in patients who will develop DCI (compared to non-DCI patients), provide further evidence that neutrophils/NETs may be a real therapeutic target. Complete depletion of neutrophils in humans may not be a viable approach, but preventing NETosis via a more selective approach, such as PAD4 inhibition, may be especially notable as even delayed treatment was effective. Finally, neutrophils/NETs have a stronger rationale as a therapeutic target as humans have 2-3 times as many circulating neutrophils as mice.⁶⁸ More experimental studies are required to validate neutrophils and NETs as therapeutic targets to prevent DCI, but if a beneficial effect is observed, this will encourage the development of future clinical trials, particularly with NET inhibitors, that are currently being tested in humans.⁶⁹

Limitations

No dose study was performed for any intervention as they have all been utilized for the specific indications in other mice injury models. We did not test a delayed treatment regimen. As PAD4 was observed to have a remarkable treatment effect when given 3 hours post-SAH, we will examine delayed treatment of PAD4 inhibition in a future study. Another limitation is that all outcomes were examined at a maximum of 7 days post-SAH. Future studies will investigate the outcomes of mice administered these treatments at 30 days post-SAH to evaluate if treatment has long-term beneficial effects. A limitation of our intravital imaging experiment is that we used vessel markers that are flow-dependent. To confirm iNETs formation, future studies will utilize transgenic mice expressing vessel markers with fluorescent tags.

As SAH is slightly more predominant in women, we only used female mice.^26,27 Future studies need to use males and aged mice. Interestingly, neutrophil depletion is reported to be even more therapeutically beneficial in aged mice with ischemic stroke.^70,71

This study examined the localization of iNETs with delayed infarcts, but further analysis remains. Analysis of infarcts identified via Nissl staining showed that the majority of infarcts left tissue voids rather than stainable tissue. However, in 7 SAH mice that displayed an area of neuronal loss (i.e. infarct), we observed iNETs clusters. Figure S12 shows a SAH mouse with infarction on the left hemisphere (lack of tissue not caused by mechanical issues) and an infarct in the right hippocampal region. NETs staining shows an iNETs cluster in the right hippocampal region located within the infarct. Since only 7 SAH mice examined had stainable infarcts and iNETs clusters were located in 6/7 mice, the data indicates that there is potentially a localization of iNETs within infarcted tissue. Additional examination is required to determine if iNETs localize to infarcts. Also, NETs counting was performed in a single 40 μm thick slice from each brain. There is a chance that counting in multiple slices throughout the brains for each animals gives a better understanding of the brain iNETs burden.

Since this study focused on iNETs as a cause of DCI and poor outcome after SAH, we did not examine the mechanism by which the NETs are created. To date, three major mechanisms of NETs exist which include the classical/suicidal, vital, and mitochondrial NETosis.^57,58 Future studies are needed to identify which NETosis mechanism generates the NETs after SAH. However, since PAD4 and citrullinated histones are primary components of the NETosis mechanisms,^57,58 PAD4 inhibition is likely a therapeutic target regardless of the specific NETosis mechanism.

A limitation of the human study is that we examined brain tissue 1-day post-rupture for the patient who did not develop DCI and on day 12 post-rupture for the patient who developed DCI. Ideally, the no DCI sample would be evaluated 4-14 days post-rupture.

Conclusion

Neutrophils infiltrate the brain within the first 24 hours after SAH with iNET formation as early as day 1. Depletion of neutrophils and NET inhibition mitigate neurological deficits and DCI following SAH. Blood markers of NETs are elevated in DCI patients and may be used as predictors of DCI. This study highlights that not only are neutrophils and NETs are potential therapeutic targets for SAH patients, but that inhibiting neutrophils or preventing NETs may also prevent DCI.

Highlights.

• Neutrophil extracellular traps occlude brain blood vessels after SAH

• Markers of neutrophil extracellular traps can predict which SAH patient is at-high risk for developing delayed cerebral ischemia

• Inhibition of neutrophil extracellular traps improves functional outcome and reduces delayed neurological deficits after SAH in mice

Glossary

SAH

Subarachnoid hemorrhage

NETs

Neutrophil extracellular traps

DCI

Delayed cerebral ischemia

iNETs

Intravascular brain NETs

PAD4

Peptidylarginine deiminase 4

CBF

Cerebral blood flow

DWI

Diffusion-weighted images

DND

Delayed neurological decline