Inversion of neurovascular coupling after subarachnoid hemorrhage in vivo

Abstract

Subarachnoid hemorrhage (SAH) induces acute changes in the cerebral microcirculation. Recent findings ex vivo suggest neurovascular coupling (NVC), the process that increases cerebral blood flow upon neuronal activity, is also impaired after SAH. The aim of the current study was to investigate whether this occurs also in vivo. C57BL/6 mice were subjected to either sham surgery or SAH by filament perforation. Twenty-four hours later NVC was tested by forepaw stimulation and CO₂ reactivity by inhalation of 10% CO₂. Vessel diameter was assessed in vivo by two-photon microscopy. NVC was also investigated ex vivo using brain slices. Cerebral arterioles of sham-operated mice dilated to 130% of baseline upon CO₂ inhalation or forepaw stimulation and cerebral blood flow (CBF) increased. Following SAH, however, CO₂ reactivity was completely lost and the majority of cerebral arterioles showed paradoxical constriction in vivo and ex vivo resulting in a reduced CBF response. As previous results showed intact NVC 3 h after SAH, the current findings indicate that impairment of NVC after cerebral hemorrhage occurs secondarily and is progressive. Since neuronal activity-induced vasoconstriction (inverse NVC) is likely to further aggravate SAH-induced cerebral ischemia and subsequent brain damage, inverse NVC may represent a novel therapeutic target after SAH.

Introduction

Subarachnoid hemorrhage (SAH) is a subtype of hemorrhagic stroke responsible for 6 to 10% of all stroke events.¹ Due to its high mortality rate of almost 50%, SAH accounts for about 20% of cerebrovascular related deaths.² Additionally, 30% of all SAH survivors remain severely disabled.³ In 85% of cases, SAH is caused by a spontaneous rupture of a cerebral aneurysm located at the base of the skull with subsequent bleeding into the subarachnoid space. As a consequence of hematoma formation, intracranial pressure (ICP) spikes to levels that substantially decrease cerebral perfusion pressure (CPP), causing global cerebral ischemia. Cerebral blood flow (CBF) recovers after a few minutes, only to significantly decrease again within the next few hours.⁴ Data from experimental studies directly visualizing cerebral microvessels after SAH have shown that post-hemorrhage ischemia is characterized by persistent spasms of pial arterioles which cause reductions of CBF and thrombus formation resulting in lack of microvascular perfusion.^5–7 Accordingly, basal blood flow is reduced after SAH and may result in ischemic brain damage and bad outcome.

In addition to the basal level of CBF, it is also critically important that cerebral vessels dilate upon activation of nearby neurons in order to match the local delivery of oxygen and glucose to focal increases in metabolic demand within the central nervous system. This process, called neurovascular coupling (NVC), was recognized by Moore and Cao⁸ and Roy and Sherrington⁹ and is mediated by both metabolic agents—e.g. increased CO₂ and decreased pH—as well as complex signaling between neurons, astrocytes, pericytes, and endothelial cells—the so called neurovascular unit. So far, however, very little is known about the consequences of SAH on NVC in vivo.

We recently demonstrated in vivo that within 3 h after SAH pial and parenchymal, microvessels lose their ability to respond to CO₂, while NVC is completely preserved.^10,11 However, when assessing NVC ex vivo in brain slices from SAH-operated rats, we showed a time-dependent impairment/inversion of NVC lasting for up to 96 h after SAH.¹² Since thus far it is unclear whether NVC is also impaired in vivo following SAH, the aim of the current study was to subject mice to SAH and directly visualize the response of their pial and parenchymal vessels upon forepaw stimulation and CO₂ inhalation 24 h after hemorrhage.

Materials and methods

Animal breeding, housing, and all experimental procedures were conducted according to institutional guidelines. Experiments performed at the Institute for Stroke and Dementia Research, University of Munich Medical Center were approved by the Ethical Review Board of the Government of Upper Bavaria (protocol number 136-11). Experiments performed at the Department of Pharmacology, University of Vermont were approved by the Institutional Animal Care and Use Committee at the University of Vermont. Male C57BL/6 mice (20 to 23 g body weight Charles River Laboratory, Sulzfeld, Germany and Charles River Laboratory, Saint Constant, Quebec, Canada) were used for this study. Experiments were planned, carried out, and reported according to the ARRIVE guidelines.¹³

Animal preparation and monitoring

Experimental animals had free access to food and water before and after surgery. For induction of SAH, anesthesia was induced by intraperitoneal injection of midazolam (5 mg/kg; Braun, Melsungen, Germany), fentanyl (0.05 mg/kg; Jansen-Cilag, Neuss, Germany), and medetomidine (0.5 mg/kg; Pfizer, Karlsruhe, Germany) as previously described.^14,15 Mice were orotrachealy intubated and mechanically ventilated (Minivent, Hugo Sachs, Hugstetten, Germany). End-tidal pCO₂ was measured with a microcapnometer (Capnograph, Hugo Sachs, Hugstetten, Germany) and kept constant between 30 and 40 mmHg by respective adjustments to the ventilation. To maintain body temperature at 37℃ a thermostatically regulated, feedback-controlled heating pad (FHC, Bowdoin, ME, USA) was used. ICP was measured in each animal for 15 min after SAH using a microsensor-based ICP probe (Codman & Shurteff Inc, Raynham, MA) to prove successful induction of SAH as described before.¹⁶For continuous monitoring of regional cerebral blood flow (rCBF), a flexible laser-Doppler probe (Periflux 4001 Master, Perimed, Stockholm, Sweden) was glued onto the skull above the territory of the left middle cerebral artery (MCA). Blood gases and electrolytes were measured at the end of each experiment.

For experiments on NVC and vasoreactivity, mice were initially anesthetized with 2% isoflurane in 70% N₂O and 30% O₂. Later on, isoflurane was gradually reduced over the course of 10 min to a range of 0.5 to 0.9% in 70% room air and 30% O₂. At the same time, a continuous intra-arterial infusion of ketamine (30 mg/kg/h, Inresa, Freiburg, Germany) was administered.¹⁷

Induction of SAH

SAH was induced using the filament perforation model as previously described.^16,18,19 Briefly, a 5–0 monofilament was introduced via the left external carotid artery into the internal carotid artery and advanced intracranially. SAH induction was marked by a sudden increase of ICP. Immediately after, the filament was withdrawn and the external carotid artery was ligated. The same procedure was followed in sham-operated mice with the only exception that the filament was not advanced far enough to induce hemorrhage. Anesthesia was terminated by intraperitoneal injection of atipamezole (2.5 mg/kg; Pfizer) naloxone (1.2 mg/kg; Inresa, Freiburg, Germany), and flumazenil (0.5 mg/kg; Hoffmann-La-Roche, Grenzach-Wyhlen,Germany). Thereafter, mice were kept in an incubator at 33℃ for 2.5 h. Identical surgical procedures were used for the induction of SAH in mice used for ex vivo studies with the exception that inhalation of isoflurane (4% induction; 2% maintenance) was used to achieve a surgical plane of anesthesia.

Forepaw-evoked NVC

Mice were re-anesthetized 24 h after SAH and the CBF response after NVC was evaluated as previously described.^11,17 Briefly, the left forepaw was stimulated with two subdermal needle electrodes with a diameter of 0.2 mm (Hwato, Suzhou, China) at an intensity of 2 mA for 0.3 ms (Digitimer Ltd, Hertfordshire, England). One stimulation cycle contained 96 stimulations and lasted for 16 s (6 Hz). The interval between two stimulation cycles was 40 s. CBF was assessed at five different locations covering the whole somatosensory cortex. The region with the strongest CBF response was also continuously stimulated (2 mA) for 1 min in order to evaluate the response to a tonic stimulus. This region was then considered for analysis and further for assessment of the microvascular response by two-photon microscopy (Figure 1(a)).

Figure 1.

SAH induction and experimental design. (a) Schematic representation of the experimental design for both SAH and sham surgery and forepaw stimulation followed by 10% CO₂ inhalation. (b) Intracranial pressure and (c) cerebral blood flow starting 5 min before SAH induction and continuing for the first 15 min after arterial perforation. A sudden increase in ICP and drop in CBF confirm vessel perforation. Mean ± SD; n = 11 in sham and n = 9 in SAH group.

Two-photon microscopy

After assessing the CBF response to forepaw stimulation, a cranial window (2 × 1 mm) was drilled under constant cooling above the area of the somatosensory cortex associated to the forepaw leaving the dura mater intact as previously described.²⁰ Mice were placed under a two-photon microscope (Zeiss LSM-7 MP, Oberkochen, Germany) equipped with a Li:Ti laser (Chameleon, Coherent, USA) as described previously,²¹ and the exposed dura mater was kept wet with isotonic saline. The fluorescent plasma dye, fluorescein isothiocyanate (FITC-dextran; molecular weight 150 kDa) was given systematically via femoral artery injection (0.05 ml of a 0.5% solution; Sigma, Deisenhofen, Germany) and all parenchymal (diameter: 5–20 µm; depth: 150 µm) and pial arterioles (diameter: 20 to 40 µm) in the region were visualized using two-photon fluorescence microscopy and a 10 × Zeiss EC Plan-NeoFluar objective. Pial arterioles were followed into the parenchyma along an axis normal to the brain surface. Arterioles were distinguished from venules on the basis of velocity and direction of blood flow.

Vascular reactivity to CO₂

Diameters of both parenchymal and pial arterioles were examined under physiological conditions in order to obtain baseline values. Thereafter, arteriolar diameter was observed during inhalation of 10% CO₂ for 10 min. The amount of inhaled CO₂ was measured by microcapnometry (Hugo-Sachs Elektronik, March-Hugstetten, Germany). Arteriolar diameters were quantified with calibrated image analysis software (Zen, Zeiss, Oberkochen, Germany) and expressed in percentage of baseline as previously described.¹⁷

Vessel diameter during forepaw-evoked NVC

For evaluation of changes in vessel diameter after forepaw stimulation, the same protocol was used as for assessment of CBF. Briefly, the region previously identified to yield the most pronounced CBF response was activated by forepaw stimulation using a cycle that contained 96 stimulations and lasted for 16 s (6 Hz). The interval between two stimulation cycles was 40 s. Thereafter, the same region was continuously stimulated (2 mA) for 1 min in order to evaluate the response to a tonic stimulus.

NVC ex vivo using freshly prepared brain slices

NVC was examined ex vivo in freshly prepared brain slices obtained from control and 24-h SAH animals. The diameter of parenchymal arterioles was measured by infrared-differential interference contrast (IR-DIC) microscopy as previously described.^12,22 Briefly, brains were dissected from mice after decapitation under deep pentobarbital anesthesia (60 mg/kg) and sliced into 160 µm thick cortical slices from the MCA territory using Leica VT1000S vibratome. Brain slices were perfused with artificial cerebral spinal fluid (aCSF; 125 mM NaCl, 3 mM KCl, 18 mM NaHCO₃, 1.25 mM NaH₂PO₄, 1 mM MgCl₂, 2 mM CaCl₂, 5 mM glucose and 0.4 mM ascorbic acid, aerated with 5% CO₂/20% O2/75% N2, pH 7.35, 35–37℃) throughout the experiment. The thromboxane A₂ analog, U46619 (125 nM, Calbiochem), was added to aCSF to establish a similar level of arteriolar tone in slices from both animal groups. Arteriolar diameter before EFS were 4.28 ± 0.35 µm (control) and 4.59 ± 0.20 µm (SAH 24 h), and not significantly different between groups. NVC was initiated using electrical field stimulation (EFS, 50 Hz, 0.3-ms alternating square pulse, 3-s duration). IR-DIC images (820 nm) were acquired at ∼ 1 Hz using a Biorad Radiance multiphoton imaging system, and arteriolar diameter was measured at three points along the segment (∼10 µm) of the arteriole exhibiting the largest response to EFS by SparkAn image analysis software (written by Dr. Bonev, Univeristy of Vermont).

Statistical analysis

For in vivo studies, statistical analysis was performed with a standard statistical software package (Sigma Plot 12.5; Systat Software, Erkrath, Germany). Results for CO₂ reactivity experiments are presented as mean ±standard error of the mean (SEM). Results for somatosensory stimulation are presented as median ± 75/25 percentile. Differences across groups were evaluated using the Mann–Whitney Rank Sum test with the Bonferroni correction. For the ex vivo study, Origin 9.1 software (Origin Lab, Northampton, MA, USA) was used for statistical analysis. EFS-induced changes in arteriolar diameter are expressed as percent change in diameter (mean ± SEM) and were evaluated by unpaired t-test.

Randomization and blinding

All animals were randomly assigned to the procedures; the surgical preparation and data analysis were performed by a researcher blinded towards the treatment of the animals.

Results

In vivo physiological parameters and mortality

All physiological parameters—body temperature, systemic blood pressure, blood pH, pCO₂ and pO₂—factors shown to effect CBF²³ were carefully monitored during the procedure and did not differ between groups (Supplementary table 1). When SAH was induced, the resulting increased ICP caused a reduction in CPP and a reduction of CBF of approximately 85%. After ICP decreased to values around 30 mmHg, CBF stabilized and remained constant for the remainder of the observation period (Figure 1(b) to (c)). Two mice from each group (4 out of 21) died before imaging could be performed, and another mouse died during procedures after CBF recording.

In vivo CO₂ reactivity in cerebral pial arteries and parenchymal arterioles is abolished 24 h following SAH

Direct visualization of pial and parenchymal vessels with two-photon microscopy showed a compromised endothelium-dependent response after SAH. In sham-operated mice, pial arterioles dilated upon inhalation of 10% CO₂ (Figure 2(a) and (b) white symbols). After SAH, however, arteries constricted rather than dilated in response to CO₂ inhalation (Figure 2(a) and (b), gray symbols). Considering the important role of the microcirculation in the maintenance of CBF, we also investigated CO₂ responsiveness of parenchymal arterioles. At a depth of about 150 µm into the somatosensory cortex arterioles in the sham-operated mice dilated by >20% in response to inhalation of 10% CO₂ (Figure 2(a) and (c) white symbols). Following SAH parenchymal arterioles showed only an initial increase in diameter of about 10% in response to inhaled CO₂ and then failed maintaining this response over time. We even observed a gradual decline in dilation until vessel diameter returned to baseline values despite continued inhalation of CO₂ (Figure 2(d) and (e), gray symbols). These results demonstrate that the vasodilation observed in healthy mice in response to CO₂ exposure is dramatically abolished in pial and parenchymal arterioles after SAH. Moreover, these data indicate that the severe endothelium-dependent dysfunction that we previously observed 3 h after SAH ¹¹ remains 24 h after the initial bleeding.

Figure 2.

Pial and parenchymal arteries do not respond to hypercapnia 24 h after SAH. (a) Representative two-photon microscopy images of pial (top) and parenchymal arterioles (bottom) of SAH and sham-operated mice before and during 10% CO₂ inhalation. (b) Surface and (d) parenchymal artery diameter during hypercapnia of mice subjected to sham surgery (white symbols) or SAH (gray symbols). Pial and parenchymal vessel dilated normally in response to inhalation of 10% CO₂ in sham-operated mice (white symbols), while no response was observed after SAH (gray symbols). (c–e) Box plots show the quantification of the integral artery diameter that corresponds to the area under the curve.

Mean ± SEM; Mann–Whitney Rank Sum test (P < 0.05 vs. 0 min and P < 0.01 vs. SAH). Median ± 75/25 percentile; Mann–Whitney Rank Sum test (P < 0.001). 15 to 25 arteries (b–c) and 18 to 26 arteries (d–e) in n = 7 to 8 mice per group.

In vivo NVC is compromised 24 hours after SAH

Twenty-four hours after hemorrhage, the increase of CBF in response to repetitive sensory stimulation of the forepaw was comparable between sham-operated and SAH mice (Figure 3(a) to (c)). A stronger, continuous sensory stimulation of the forepaw for 1 min, however, increased CBF only in sham-operated mice (Figure 3(d) to (e), white symbols), while the response in SAH mice was significantly attenuated (Figure 3(d) to (e), gray symbols).

Figure 3.

Changes in CBF in response to neurovascular coupling 24 h after SAH. (a) Sensory stimulation to 10 consecutive stimuli shows an increase in CBF response that attenuates in SAH mice until it becomes statistically significant from the response in sham mice at the 10th stimulus. (b–c) Box plots showing CBF increase in response to the first four (b) and last six (c) discrete electrical stimuli to the forepaw, in sham-operated mice (white symbols) and after SAH (gray symbols). No significant effect was found between the experimental groups. (d–e) CBF increase in response to continuous electrical stimulation shows a significant difference between sham-operated mice and mice subjected to SAH, though this is not apparent when quantifying the area under the curve (e). Mean ± SEM; Mann–Whitney Rank Sum test (P < 0.05 vs. 0 s and P < 0.01 vs. SAH). Median ± 75/25 percentile. n = 7 to 9 per group.

Visualization of parenchymal arterioles of up to 20 µm in diameter in response to repetitive forepaw stimulation showed a similar picture. Stimulation caused an increase in parenchymal arteriolar diameter in the sham-operated group of about 20% (Figure 4(a) to (c), white symbols). After SAH, this response was absent; even a tendency towards inversion of NVC was observed (Figure 4(a) to (c), gray symbols). When using a stronger stimulation paradigm, i.e. continuous forepaw stimulation for 1 min, this response became more evident. Arterioles dilated by 25% in sham-operated animals (Figure 4(d) and (e), white symbols), but massive and significant arterial constriction was observed in SAH mice (Figure 4(d) and (e), gray symbols). These results demonstrate a severe dysfunction of NVC 24 h after SAH in vivo.

Figure 4.

Two-photon imaging of arteriolar diameter demonstrates severe impairment of neurovascular coupling 24 h after SAH. (a) Representative two-photon microscopy image of parenchymal vessel in sham-operated mice (top) and SAH mice (bottom) at baseline and after stimulation. (b–c) Arteriolar dilation in response to a discrete electrical stimulation shows a significant difference between experimental groups in the mean change in diameter (b) and the quantification of the integral of arteriolar diameter (c). (d–e) Arteriole dilation in response to a continuous electrical stimulation shows a significant difference between SAH or sham-operated mice. Mean ± SEM; Mann–Whitney Rank Sum test (P < 0.05 vs. 0 min and P < 0.01 vs. SAH). Median ± 75/25 percentile; Mann–Whitney Rank Sum test. 27 arterioles (b–c) 18 to 22 arterioles (D-C) in n = 7 to 8 mice per group.

Inversion of NVC in brain slices from 24-hour SAH animals

We have previously reported that ex vivo NVC was preserved (i.e. EFS-induced vasodilation) in the acute phase (first 3 to 4 h) of SAH in both rat²² and mouse SAH model animals.¹¹ Here, we have examined ex vivo NVC in brain slices obtained from 24-h SAH model mice. As previously shown, focal neuronal activation by EFS led to arteriolar dilation in brain slices from un-operated control animals (20.8 ± 1.7% increase in diameter, n = 6). By contrast, EFS caused arteriolar constriction in the majority of brain slices from 24-h SAH animals (five out six arterioles, ∼83%). On average, EFS caused a constriction representing a decrease in diameter of 16.1 ± 6.1% (n = 6) in brain slices from the 24-h SAH group that was significantly different from the dilation observed in the control group (Figure 5(a)). Individual arteriolar responses to EFS are shown in Figure 5(b). Consistent with in vivo observations (Figure 4), these results demonstrate ex vivo inversion of NVC 24 h after SAH.

Figure 5.

EFS-induced vasoconstriction in brain slices 24 h after SAH. (a) Infrared-differential interference contrast (IR-DIC) images of cortical brain slices from control and 24 h SAH mice, showing inversion of neurovascular coupling 24 h after SAH. EFS induced vasodilation in brain slices from control animals (six out of six arterioles). However, in brain slices obtained from animals 24 h after SAH, EFS caused marked vasoconstriction in five out of six brain slices. Dashes outline the intraluminal diameter of parenchymal arterioles. Scale bars, 10 µm. (b) Summary of EFS-evoked changes in arteriolar diameter in brain slices from control and 24 h SAH animals (n = 6 each). Opened (control) and closed (24 h SAH) circles represent individual data points. Bar graph expresses mean percent change in diameter ± SEM. ** P < 0.01, unpaired t-test.

Discussion

Based on our previous work investigating NVC within the first few hours after SAH, the aim of our current investigation was to examine how in vivo and ex vivo neurovascular communication evolves over time after SAH. In order to do so, we used two-photon microscopy to visualize reactivity of both pial and parenchymal vessels 24 h after induction of SAH in mice. Our data show that after SAH, cerebral vessels across the vascular tree—pial and parenchymal arteries—fail to dilate in response to CO₂, which is a vasoactive metabolic byproduct that specifically dilates cerebral vessels.^24,25 Moreover, our data show a delayed dysfunction in NVC in vivo as well as ex vivo, with an inverse response 24 h after SAH, indicating that after SAH, neuronal activity constricts rather than dilates cerebral vessels.

Severe endothelium-dependent dysfunction in vivo was identified for the first time by direct observation of pial vessels 3 and 24 h after SAH using conventional epi-fluorescence microscopy.^10,26,27 However, due to the limited penetration depth of the technology used in these studies, the investigation was limited to vessels on the brain surface. This limitation was overcome using two-photon microscopy, to provide the first in vivo measurements of parenchymal arteriolar functionality 3 h after SAH.¹¹ It was found that in addition to microvessels located on the brain surface, arterioles in the parenchyma lost their ability to dilate in response to CO₂ inhalation. This study investigated the very acute phase after hemorrhage (i.e. 3 h) when vasospasm occurs in microvessels; however, it remained unclear if this process was short-lived or if it persisted beyond the acute phase of SAH. In the current study, we show that this early loss of CO₂ reactivity in pial and parenchymal vessels persists for at least 24 h after hemorrhage. The relevance of this finding is that following SAH, the brain seems to be unable to adjust blood flow to neuronal activity. Hence, each firing of neurons causes a mismatch of flow and metabolism and may therefore further damage the brain. Mechanistically, the results of the current study suggest that endothelial NO signaling may be dysfunctional after SAH. This is based on the observation that CO₂ reactivity is at least partially dependent on constitutive endothelial NOS, findings consistent with studies showing that during SAH, NO concentration, and eNOS activity drop with the onset of hemorrhage.^28,29

Importantly, we also show that following hemorrhage, the mechanisms of communication between neurons and vessels that in healthy conditions lead to a balanced supply of oxygen and nutrients to the brain via the blood stream are severely altered 24 h after SAH. Recent ex vivo studies investigating intra-parenchymal arterioles in rats after SAH reported an inversion of NVC: arterioles constricted rather than dilated in response to neuronal activity 24 to 96 h after SAH.³⁰ While our previous study investigating neurovascular reactivity showed that NVC is not impaired within the first few hours after SAH,¹¹ in the current study, we unequivocally demonstrate in vivo and ex vivo a progression in vascular dysfunction, affecting NVC 24 h after SAH. These findings define a therapeutic time window that could lead to an improvement in the outcome of already promising approaches like NO-based strategies, i.e. administration of inhaled NO²⁰ or NO donors.³¹

From a mechanistic point of view, recent studies in rat brain slices have provided evidence for a link between SAH-induced inversion of NVC from vasodilation to vasoconstriction and high-amplitude spontaneous calcium (Ca²⁺) events within astrocytic end-feet, shown to occur only in the end-feet surrounding arterioles that exhibited an inverted response. These changes in signaling have been shown to be causal to the switch in polarity of NVC after SAH.²² Additionally, the percentage of these end-foot high-amplitude Ca²⁺ signals paralleled the development of the switch in polarity in a time-dependent manner, further corroborating the existence of a therapeutic time window. Broad-spectrum inhibition of purinergic (P2) receptors and targeted inhibition of G_q-coupled P2Y receptors have been shown to abolish high-amplitude Ca²⁺ signals after SAH, but the purine nucleotides responsible were shown not to be released by neurotransmission after SAH.³² This suggests that the therapeutic targeting of purinergic signaling in astrocytes could be complementary to an early rescue of NO reactivity.

In summary, the current study demonstrates that pial and parenchymal cerebral vessels lose their ability to respond to metabolic stimuli like CO₂ after hemorrhage. Further on, NVC is inversed 24 h after SAH, i.e. neuronal activation results in vasoconstriction instead of vasodilatation. Together with our previous findings showing no impairment of NVC 3 h after SAH, the current results indicate progressive vascular dysfunction following cerebral hemorrhage. Accordingly, the current study provides the first information on the functional integrity of parenchymal microvessels after SAH in vivo. The clinical relevance of our results is that any neuronal activation, e.g. by sensory stimulation, may result in vasoconstriction, mismatch of metabolism and blood flow, relative ischemia, and further damage in SAH patients. Accordingly, SAH patients may need to be handled with special attention until proper neurovascular function is restored or respective therapies are developed to prevent post-hemorrhagic dysfunction of cerebral vessels.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Solorz-Żak foundation, by the American Heart Association (14SDG20150027), National Institutes of Health (P01 HL095488, P30 RR032135, P30 GM103498 and S10 OD10583), Totman Medical Research Trust Fund, the Peter Martin Brain Aneurysm Endowment, and the Deutsche Forschungsgemeinschaft through the Munich Cluster of Systems Neurology (Synergy).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

MB: performed all in vivo experiments and wrote the manuscript. MK: performed all ex vivo experiments and wrote the manuscript. GCW: designed the study, wrote and edited the manuscript. NP: designed the study, wrote and edited the manuscript.

Supplementary material

Supplementary material for this paper can be found at http://jcbfm.sagepub.com/doi/suppl/10.1177/0271678X16686595