Mitochondrial dysfunction associated with autophagy and mitophagy in cerebrospinal fluid cells of patients with delayed cerebral ischemia following subarachnoid hemorrhage

Decreased mitochondrial membrane potential in cerebrospinal fluid (CSF) was observed in patients with subarachnoid hemorrhage (SAH) accompanied by delayed cerebral ischemia (DCI). However, whether abnormal mechanisms of mitochondria are associated with the development of DCI has not been reported yet. Under cerebral ischemia, mitochondria can transfer into the extracellular space. Mitochondrial dysfunction can aggravate neurologic complications. The objective of this study was to evaluate whether mitochondrial dysfunction might be associated with autophagy and mitophagy in CSF cells to provide possible insight into DCI pathogenesis. CSF samples were collected from 56 SAH patients (DCI, n = 21; and non-DCI, n = 35). We analyzed CSF cells using autophagy and mitophagy markers (DAPK1, BNIP3L, BAX, PINK1, ULK1, and NDP52) via qRT-PCR and western blotting of proteins (BECN1, LC3, and p62). Confocal microscopy and immunogold staining were performed to demonstrate the differentially expression of markers within dysfunctional mitochondria. Significant induction of autophagic flux with accumulation of autophagic vacuoles, increased expression of BECN1, LC3-II, and p62 degradation were observed during DCI. Compared to non-DCI patients, DCI patients showed significantly increased mRNA expression levels (2−ΔCt) of DAPK1, BNIP3L, and PINK1, but not BAX, ULK1, or NDP52. Multivariable logistic regression analysis revealed that Hunt and Hess grade ≥ IV (p = 0.023), DAPK1 (p = 0.003), and BNIP3L (p = 0.039) were related to DCI. Increased mitochondrial dysfunction associated with autophagy and mitophagy could play an important role in DCI pathogenesis.


Results
Clinical outcomes. A total of 56 SAH patients treated with endovascular coil embolization were included in the analysis (Table 1). Twenty-one (37.5%) patients experienced DCI during the follow-up. The Cohen's kappa score for DCI diagnosis was 0.889, indicating almost perfect agreement (Supplemental Data). There was no statistical difference in clinical findings (e.g., female gender, age, hypertension, diabetes mellitus, hyperlipidemia and smoking status) between the group with DCI diagnosis and the group without DCI diagnosis. Hunt and Hess grade ≥ IV was observed in 11 (31.4%) patients without DCI and 11 (52.4%) patients with DCI. Poor outcome was observed in 20 (35.7%) patients. The number of anterior circulation aneurysms in DCI and non-DCI patients were 17 (81.0%) and 28 (80.0%), respectively.
Morphological changes in mitochondria of SAH patients with DCI. Transmission electron microscopy (TEM) was used to investigate the changes in subcellular ultrastructure of CSF cells obtained from SAH patients with DCI (Fig. 1). Numerous autophagic vacuoles containing mitochondria and fusion of autophagic vacuoles with abnormal mitochondria with swollen matrix and collapsed cristae accumulated in CSF cells. Additional autophagic vacuoles appeared close to the swollen mitochondria with inner mitochondrial membrane, indicating that the mitochondrial dysfunction with morphological impairment was associated with autophagic flux, and suggesting the possibility of autophagy and mitophagy in CSF cells in DCI pathogenesis.

Increased autophagy in mitochondria of CSF cells. Quantitative real-time PCR (qRT-PCR) analysis
was performed to evaluate mRNA levels of autophagy and mitophagy markers in CSF cells of SAH patients with DCI compared to those in CSF cells of SAH patients without DCI. DCI patients exhibited significantly higher expression (2 −ΔCt ) than non-DCI patients: death-associated protein kinase ( Fig. 2A). We further analyzed autophagy markers including ULK1, BECN, LC3, and p62 in CSF cells of DCI (n = 6) versus non-DCI patients (n = 6). Western blot revealed increased protein expression of BECN1 and its phosphorylation at Ser15 in DCI patients compared with non-DCI patients ( Fig. 2B, C, Supplemental Figure S1, and Supplemental Table S2). In addition, increased LC3II expression and p62 degradation were also observed in DCI patients.

Autophagy and mitophagy in vWF-positive CSF cells in DCI.
To investigate whether mitochondrial dysfunction with increased autophagy and mitophagy occurred in vWF-positive CSF cells in SAH patients with DCI, multi-colored immunofluorescence staining was conducted with antibodies specific for vWF and MTDR as a marker of dysfunctional mitochondria (Fig. 3). The fluorescence signals of DAPK1, BNIP3L/NIX, PINK1, and BECN1 overlapped with that of Mito Tracker Deep Red (MTDR) in vWF-positive CSF cells. vWF-positive cells (green) were not stained with CD41 (platelet marker). The results suggested that increased mitochondrial dysfunction with autophagy and mitophagy in vWF-positive CSF cells are associated with DCI pathogenesis in SAH patients.

Mitochondrial dysfunction concomitant with increased DAPK1 in DCI.
We further performed immunogold labeling with anti-DAPK1 to confirm the subcellular localization of enhanced DPAK1 in CSF cells obtained from SAH patients with DCI (Fig. 4). Interestingly, DAPK immunogold particles were mainly observed in damaged mitochondria with abnormal morphology, which is consistent with results presented in Fig. 3. Taken www.nature.com/scientificreports/ together, these results suggest that increased DAPK1 may play a pivotal role in mitochondrial dysfunction during the development of DCI.

Discussion
The clinical significance of mitochondrial dysfunction has not been well studied in the DCI following SAH. In the event of cerebral ischemia, mitochondria are damaged through a series of following steps: (1) depolarization of mitochondrial membrane potential; (2) PINK2 accumulation with decreased adenosine triphosphate (ATP) reduction; (3) abnormal mitochondrial fission and fusion; and (4) broken cellular homeostasis due to increased ROS and matric calcium and subsequent cell death 12 . Autophagy and mitophagy are induced to remove damaged mitochondria after an ischemic insult. However, results did not always show only neuroprotection 12 .
Baek et al. 16 reported that attenuated autophagic signaling by carnosine, an endogenous pleiotropic peptide, was associated with improvement of mitochondrial function against ischemic brain injury. On the other hand, BNIP3L causes excessive mitophagy leading to delayed neuronal loss 17 . Chen et al. 18 reported that -(-) Epigallocatechin-3-Gallate (EGCG) administration was beneficial for maintain autophagy flux by regulating Beclin-1 and LC3B. EGCG decreased oxyhemoglobin-induced mitochondrial dysfunction with decreased ROS production following SAH. Cao et al. 19 also reported that melatonin lessened mitophagy-associated NLRP3 inflammasome for EBI after SAH. Nonetheless, the therapeutic role of mitophagy in the DCI is still unknown. Compared with EBI, which occurs in the brain parenchyma in the first 72 h after SAH onset, DCI usually develops 3 days thereafter and peaks from days 7 to 14. SAH literally refers to acute hemorrhage within the subarachnoid space surrounding the cerebral arteries. Free hemoglobin within the subarachnoid space triggers oxidative stress and inflammation, resulting in DCI 20 . Chou et al. 14 initially investigated the functional relevance of mitochondria in human CSF samples following SAH. In their study, higher mitochondrial membrane potential reflected favorable functional outcomes during the 3-month follow-up. In particular, the higher percentage of mitochondria potentially originating in astrocytes was associated with a favorable outcome. In this study, we further evaluated autophagy and mitophagy markers in the CSF cells to reveal the underlying DCI pathogenesis based on mitochondrial dysfunction. Our study showed that increased mitochondrial dysfunction associated with autophagy and mitophagy in CSF cells may drive DCI pathogenesis. Among the markers, the DAPK1 expression varied most significantly between the DCI and non-DCI patients. DAPK1 acts as a sensor of mitochondrial membrane potential in mitochondrial toxin-induced cell death 21 . Mitochondrial toxins, such as rotenone, and carbonyl cyanide m-chlorophenylhydrazone (CCCP), which is an uncoupler of mitochondrial oxidative phosphorylation induce loss of membrane potential and mitochondrial swelling leading to cell death via DAPK1 activation in neuroblastoma cells. In damaged mitochondria, the accumulation of PINK1 and LC3 on the outer membrane is accompanied by loss of membrane potential 22 . Strokeinduced neuronal cell death is related to DAPK1 signaling mediated via DAPK1-N-methyl-D-aspartate receptors (NMDARs), DAPK1-p53 and DAPK1-Tau in cerebral ischemia 23 . Tu et al. 24 reported that DAPK1 directly binds with NR2B subunit of NMDARs in the cortex of mice, leading to aggravation of calcium reflux by enhancing the NR1/NR2B channel conductance. In their study, the genetic deletion of DPAPK1 or administration of NR2B CT , defined as carboxyl tail region consisting of amino acids 1292-1304, inhibited calcium reflux and protected neuronal injury against ischemic insults 23,24 . DAPK-1 was also involved in the regulation of inflammation. Activation of DAPK led to decreased T cell activation and IL-2 production 25,26 . Chuang et al. 26 reported that DAPK was a regulator of T cell receptor (TCR)-activated NF-kappa B and T lymphocyte activation. Therefore, a further study elucidating the role of DAPK1 and T cell immunity in DCI pathogenesis in the CSF space is required.
In this study, we further evaluated the role of confocal microscopy to confirm the differential expression of autophagy and mitophagy markers in vWF positive-CSF cells of SAH patients with DCI 14 . The findings demonstrated colocalization with DAPK1, BNIP3L, PINK1, and BECN1-positive autophagosomes in mitochondria, suggesting mitochondrial dysfunction with autophagy and mitophagy in DCI pathogenesis. Endothelial dysfunction is related to cerebrovascular disease. Nitric oxide and endothelin released from endothelium regulates vascular function 27 . Compared with healthy controls, SAH patients exhibit elevated synthesis of endothelial microparticles 28 . In addition, increased levels of CSF and plasma endothelin were observed in SAH patients with vasospasm 27 . In DCI patients, a higher percentage of vWF-positive mitochondria was observed compared with Table 2. Multivariable logistic regression analysis for the development of delayed cerebral ischemia following subarachnoid hemorrhage. BAX Bcl-1 antagonist X, BNIP3L BCL2 interacting protein 3 like, DAPK-1 deathassociated protein kinase-1, NDP52 nuclear dot protein 52, PINK1 PTEN-induced kinase 1. www.nature.com/scientificreports/ non-DCI patients 15 . Accordingly, future studies should focus on the alleviation of the mitochondrial dysfunction in CSF cells of endothelial origin in DCI patients. This study has some limitations. First, we enrolled SAH patients treated with coil embolization, but not surgical clipping. The penetration of laminar terminalis or brain retraction injury during the clipping can contaminate the CSF component, and therefore, bias the results of CSF mitochondria. Therefore, we excluded them from our analysis. We did not include a control group either. Thus, a follow-up study including SAH patients who underwent clipping and a control group should be performed in the future. Second, no medical treatment or additional chemical angioplasty for DCI severity was considered in the interpretation of results. In general, suppression of autophagy contributes to increased cell death, but can lead to cytotoxicity under some circumstances 29 . Mitochondrial dysfunction is likely to vary depending on the initial SAH severity and thus may be associated with the patient's outcome. Although there was no statistical significance of autophagy and mitophagy markers according to the SAH severity and patient's outcome (Supplemental Table S3 and S4), additional analysis should be performed using a large number of the SAH patients. Nevertheless, our study www.nature.com/scientificreports/ represents the first investigation of mitochondrial dysfunction associated with autophagy and mitophagy in CSF cells derived from SAH patients with DCI.

Conclusions
Increased mitochondrial dysfunction with autophagy and mitophagy could play an important role in DCI pathogenesis of patients with SAH.

Materials and methods
Study population. The derivation cohort was derived from the regional stroke database between March 2016 and May 2020. We selected SAH patients from this database based on the following inclusion criteria: (1) adult patients > 18 years old; (2) SAH due to ruptured aneurysm; (3) dense localized clot and/or vertical layer of blood greater than 1 mm in thickness on computed tomography (CT); and (4) SAH patients who were treated with endovascular coil embolization. The exclusion criteria were: (1) non-aneurysmal SAH such as trauma, infection or perimesencephalic SAH, (2) patients treated with surgical clipping and (3) previous history of central nervous system disorder or mitochondrial diseases 30 . TEM was used to detect the autophagic vacuoles and morphological changes of mitochondria in CSF cells derived from SAH patients with DCI. We evaluated and compared the autophagy and mitophagy biomarkers in CSF cells of SAH patients with and without DCI using qRT-PCR. The markers included DAPK-1, BNIP3L, www.nature.com/scientificreports/ BAX, PINK1, ULK1 and NDP52. We further evaluated the expression of autophagy executor gene of BECN1 and autophagy adaptor protein of p62 31,32 . Confocal microscopy was used to identify colocalization of differentially expressed genes in vWF-positive CSF cells, which represented endothelial cell origin, and were increased in SAH patients with DCI 15 . In our study, the diagnosis of DCI wad performed through the following criteria: (1) new developed neurological changes such as motor weakness, dysphasia, and sensory change; (2) decrease consciousness by more than 2 points via the Glasgow Coma Scale score or National Institutes of Health stroke scale; (3) fluctuation of symptoms lasing more than 1 h; (4) cerebral infarction identified on CT or MRI, but not complications related to the endovascular coil embolization; (5) concomitant severe cerebral vasospasm with narrowing more than 50% compared to the initial radiological tests; and (6) excluding other causes that may neurological changes such as re-bleeding, hydrocephalus, seizures or electrolyte imbalances 2,33 . When DCI was suspected, treatment to increase blood pressure was given first. Blood pressure was targeted with the mean systolic blood pressure 20 mm Hg above the baseline, up to 200 mm Hg while looking at the clinical response. Intra-arterial chemical angioplasty using nimodipine was performed when DCI-related clinical symptoms worsened or severe vasospasm was suspected based on a TCD greater than 200 cm/s in the middle cerebral artery or 85 cm/s in the basilar 34 , despite treatment to induce hypertension. The procedure was carried out once or twice a day from 1 to 7 days after identification of DCI. Poor outcome was defined when a modified Rankin scale (mRS) score was three or more at 3-month follow-up.
After coil embolization, continuous lumbar drainage of CSF was maintained in the neurointensive care unit in every SAH patient for 1 week after ictus in our institution. Because difference in mitochondrial membrane potential of the CSF cells was most prominent on day 5 post ictus in SAH patients with and without DCI 15 , we analyzed the CSF samples obtained from days 5 to 7.
Clinical, laboratory, and radiological information was reviewed by the two investigator independently. Disagreements were resolved by the third investigator. The protocol of DCI diagnosis among reviewers are presented in detail in the Supplemental Data. Sample collection and study design were performed according to the principles of the Declaration of Helsinki and were approved by the Institutional Review Board of the Chuncheon Sacred Heart Hospital (No. 2017-9, 2018-6, and 2019-6). All methods were performed in accordance with the relevant guidelines and regulations in manuscript. Informed consent was received from the patients or their relatives.
Transmission electron microscopy. Previously, SAH patients with DCI showed depolarization of mitochondrial membrane potential, which triggers alteration in mitochondrial function and morphologies 15 . In the present study, TEM was used to investigate the changes in subcellular ultrastructure of CSF cells in SAH patients with DCI. CSF samples were centrifuged at 4000 rpm for 10 min, and the pellets were analyzed by electron microscopy 15 . The pellets were fixed overnight in 2% glutaraldehyde in cacodylate buffer (0.1 M sodium cacodylate, 2 mM MgCl2) at 4 ℃. After washing three times with cacodylate buffer at 4 ℃, the samples were postfixed in 2% osmium tetroxide for 1 h at 4 ℃. The samples were rinsed with deionized water and dehydrated through a gradient series of ethanol, ranging from 50 to 100% ethanol, 20 min each step. The samples were incubated with progressively concentrated propylene oxide dissolved in ethanol followed by infiltration with increasing concentration of Eponate 812 resin. Samples were baked in a 65 °C oven overnight and sectioned using an Ultra microtome. Sections were viewed with a Field Emission TEM unit (JEM-2100F, JEOL) at the Korean Basic Science Institute, Chuncheon, South Korea 15 .
Real time qRT-PCR. The loss of mitochondrial membrane potential leading to mitochondrial dysfunction results in activation of mitochondrial autophagy. Therefore, autophagy and mitophagy markers such as DAPK1, PINK1, BAX, BNIP3L, and NDP52 were examined in CSF cells 21,22 . Total RNAs were isolated from the CSF cells using TriZOL (Ivitrogen, USA) according to the manufacturer's instructions. The cDNA was synthesized from 5 μg of RNA using Maxime RT PreMix Kit (iNtRON Biotechnology, Korea). The expression of mitophagy-related genes was measured by qRT-PCR using the 2 × Rotor-Gene SYBR Green PCR Master Mix (Qiagen, Carlsbad, CA, USA) in the Rotor-Gene Q (Qiagen, USA). Primer sequences are presented in the Supplemental Table S1.
Immunogold staining. We further performed immunogold staining to confirm the location of the most significant differentially expressed markers within the mitochondria of CSF cells in SAH patients with DCI. CSF cells were fixed with 0.1% glutaraldehyde and 2% paraformaldehyde in phosphate buffer at pH 7. Statistical analysis. Continuous variables are expressed as the mean and ± standard deviation (SD). A chisquare or Student's t test was carried out to identify meaningful differences between DCI and non-DCI patients. Comparative analysis via qRT-PCR was performed using the Mann-Whitney U test. The results were presented as the median and 25th-75th percentiles. Quantification of western blots using the relative optical densities with actin protein as the reference and presented as the mean ± standard error of the mean (SEM). Univariate analysis was performed to find relevant factors for the development of DCI. A multivariable logistic regression analysis was then performed to identify independent variables associated with the development of DCI using variables showing p value less than 0.20 in the univariate analysis. Statistical analysis was performed with SPSS V.21 (SPSS, Illinois, USA) and GraphPad Prism software (v.6.01; GraphPad Software Inc., San Diego, CA, USA) with a statistical significance indicated at p < 0.05.
Ethical approval. Sample collection and study design were performed according to the principles of the Declaration of Helsinki and were approved by Coordinating Ethnics Committee of the Chuncheon Sacred Heart Hospital.

Data availability
Data are available from the corresponding author (JPJ) upon ethical approval from the IRB of the participating hospital.