Ketamine improves neuronal recovery following spreading depolarization in peri-infarct tissues

Abstract

Spreading depolarization (SD) has emerged as an important contributor to the enlargement of acute brain injuries. We previously showed that the NMDA receptor antagonist ketamine was able to prevent deleterious consequences of SD in brain slices, under conditions of metabolic compromise. The current study aimed to extend these observations into an in vivo stroke model, to test whether gradients of metabolic capacity lead to differential accumulation of calcium (Ca²⁺) following SD. In addition, we tested whether ketamine protects vulnerable tissue while allowing SD to propagate through surrounding undamaged tissue. Focal lesions were generated using a distal middle cerebral artery occlusion in mice, and clusters of SD were generated at 20 min intervals with remote microinjection of KCl. SDs invading into peri-infarct regions had significantly different consequences, depending on the distance from the infarct core. Proximal to the lesion, Ca²⁺ transients were extended, as compared with responses in better-perfused tissue more remote from the lesion. Extracellular potential shifts were also longer and hyperemia responses were reduced in proximal regions following SDs. Consistent with in vitro studies, ketamine, at concentrations that did not abolish the propagation of SD, reduced the accumulation of intracellular Ca²⁺ in remote regions following an SD wave. These findings suggest that deleterious consequences of SD can be targeted in vivo, without requiring outright block of SD initiation and propagation.

Graphical Abstract Text:

We examined spreading depolarization (SD) waves in a mouse stroke model. SDs were initiated by focal KCl application and propagated through a region of graded perfusion deficit created by distal middle cerebral artery occlusion. Longer lasting depolarizations (DC) and neuronal Ca²⁺ transients occurred at locations with larger perfusion deficits (proximal to occlusion), as compared with remote recording sites. Ketamine, at concentrations that did not abolish the propagation of SD, improved recovery from SD and reduced Ca²⁺ loading. These findings suggest that deleterious consequences of SD can be targeted in vivo, without requiring outright block of SD initiation and propagation.

INTRODUCTION

Spreading depolarization (SD) is a slowly propagating wave of near-complete neuroglia depolarization that has been detected in patients suffering from acute brain injuries (Dreier, Woitzik et al. 2006, Hartings, Bullock et al. 2011), including ischemic stroke (Dohmen, Sakowitz et al. 2008). SDs are exceptionally challenging to the metabolically compromised brain and contribute to secondary lesion expansion and poor neurological outcomes (Hartings, Rolli et al. 2003, Dreier, Fabricius et al. 2017, Hartings, Shuttleworth et al. 2017). Importantly, SDs can occur for days after the initial infarction (Dohmen, Sakowitz et al. 2008, Nakamura, Strong et al. 2010, Hartings, Andaluz et al. 2020, Dreier, Winkler et al. 2022) and during this time there are no clinically approved neuroprotective interventions (Powers, Rabinstein et al. 2018). Targeting SD waves may therefore provide a therapeutic opportunity to interrupt mechanisms of injury progression in ischemic stroke and other acute brain injuries.

Ketamine is a use-dependent N-methyl-D-aspartate receptor (NMDAR) antagonist that can completely block or reduce SD frequency when used at a range of sedative doses in intensive care unit (ICU) patients (Sakowitz, Kiening et al. 2009, Hertle, Dreier et al. 2012, Carlson, Abbas et al. 2018, Gregers, Mikkelsen et al. 2020, Helbok, Hartings et al. 2020). A prospective clinical trial is underway to determine whether reducing the number of SDs improves neurological outcomes (NCT05337618). SDs are initiated when a small volume of tissue becomes simultaneously depolarized (Matsuura and Bures 1971, Tang, Mendez et al. 2014). In otherwise healthy tissues, NMDAR antagonists can abolish the initiation and propagation of SD in a concentration-dependent manner (Gill, Andine et al. 1992, Peeters, Gunthorpe et al. 2007), presumably by reducing depolarizing effects of glutamate at the SD wave-front. The concentrations of glutamate achieved at the advancing wave-front are extreme (Somjen 2001), and relatively high concentrations of NMDAR antagonists are required for effective block (Pietrobon and Moskowitz 2014). The effectiveness of propagation block by NMDAR antagonists is further decreased in ischemic tissues, a feature that has been attributed to elevated extracellular potassium (K⁺) concentrations (Petzold, Windmuller et al. 2005, Reinhart and Shuttleworth 2018). High ketamine concentrations required to block SD also result in substantial sedation with associated increased risk of ICU complications (Abou-Chebl, Lin et al. 2010). However, it is possible that, even in the absence of complete blockade, ketamine can mitigate deleterious consequences of SD to provide neuroprotection and clinical utility in stroke and other brain injuries at risk of SD-mediated damage.

A second mechanism of ketamine is the reduction of the duration of the SD event itself, as demonstrated by reduced duration of the signature slow potential change of SD (i.e., “DC shift”) (Reinhart and Shuttleworth 2018). This efficacy at reducing the duration of SD, has also been seen previously with other NMDAR antagonists (Marrannes, Willems et al. 1988, Herreras and Somjen 1993, Aiba and Shuttleworth 2012). From whole-cell recordings in murine brain slices, it was shown that a burst of spontaneous presynaptic release events (recorded as spontaneous excitatory postsynaptic potentials) occur in the late phase of SD when post-synaptic neurons remain severely depolarized (Aiba and Shuttleworth 2012). These are conditions that favor the extended opening of NMDARs and postsynaptic calcium (Ca²⁺) loading and can explain the vulnerability of metabolically-compromised neurons to SD (Aiba and Shuttleworth 2012). We recently showed that this second effect of ketamine (i.e. reduction of DC shift duration and thus reduced Ca²⁺ loading after SD) improved tissue recovery from SD in brain slices (Reinhart and Shuttleworth 2018). Similar effects were seen with memantine another NMDAR antagonist (Reinhart, Humphrey et al. 2021). The current study aimed to extend these observations into an in vivo stroke model, to test 1) whether gradients of metabolic capacity around an ischemic lesion lead to demonstrable differences in the duration of SD and associated Ca²⁺ accumulation, and 2) whether ketamine administration can improve these features of SD without outright block of the events.

We employed imaging and electrophysiology techniques in vivo and monitored vascular and neuronal responses in the neocortex during a challenging cluster of SDs in the acute period after distal middle cerebral artery occlusion (dMCAo). We observed heterogeneity in all the SD characteristics we measured depending on the distance from the occlusion site, which emphasized the extreme demand of SD clusters in metabolically vulnerable tissue. With each SD, neuronal Ca²⁺ loading became more severe in vulnerable brain areas after stroke; however, a sub-anesthetic dose of ketamine reduced the duration of SDs and delayed the progression of irrecoverable Ca²⁺ transients. These results indicate that a low ketamine concentration can be used to attenuate the burden of repetitive SDs, which may be of clinical use when extended sedation is not desired.

METHODS

Animals and Study Design

All animal procedures were performed in accordance with protocol 16-200453 approved by the UNM Health Sciences Center Institutional Animal Care and Use Committee. Animals were housed in standard ventilated cages with continuous access to food and water. C57Bl/6J mice (n = 5 mice/cage) were purchased from The Jackson Laboratory (RRID:IMSR_JAX:000664J; Bar Harbor, ME) and housed in the animal care facility (> 1 week) prior to experiments. For Ca²⁺ imaging experiments, mice expressing the floxxed calcium indicator GCaMP5G under the CAG promoter (RRID:IMSR_JAX:024477; B6;129S6-Polr2a^{tm1(CAG-GCaMP5g,-tdTomato)Tvrd}/J; from (Gee et al., 2014)) were bred with mice expressing Cre Recombinase under the CamK2a promoter (RRID:IMSR_JAX:005359; B6.Cg-Tg(Camk2a-cre)T29-1Stl/J). Offspring (maintained in standard group housing with 2-5 littermates/cage) had robust GCaMP5G expression throughout the hippocampus and neocortex (Wang et al., 2013) and were utilized in experiments. A total of 31 adult (10-20 weeks) male and female mice weighing 26.6 ± 0.8g were used for all experiments in the Results. No blinding was performed in this study; however, two experiments were conducted per day by the experimenter to interleave group comparisons on the same day (i.e., sham vs. dMCAo groups or dMCAo + vehicle vs. dMCAo + ketamine groups). No formal randomization was used but care was taken to alternate which experimental group was tested first (starting between 8 – 9 am) on a given experimental day.

An initial pilot study was conducted in n = 12 C57Bl/6J mice (data not included in Results) for the experimenter to establish surgical and experimental design protocols. Pilot experiments were likewise used to establish the following animal inclusion criteria (in addition to reaching the primary endpoint): a) the absence of hemorrhage during cautery of the dMCA, b) a perfusion deficit in the ipsilateral hemisphere following dMCAo with no deficit in sham mice (relative to the contralateral hemisphere). Additionally, it was found that success rate to the primary endpoint of animal survival after dMCAo and SD challenge (Figure 1A) was low (~50%) in the absence of additional supplementation with room air (see below). A second pilot study n = 3 C57Bl/6J mice was used to determine the ketamine dose administered (described below). In total, 56 animals were used in the present study (Results: n = 31; pilot studies: n = 15; exclusions: n = 10, based on primary endpoint and inclusion criteria, above). In addition to whole-animal exclusions, individual SDs within a single animal were excluded based on outlier tests (see Statistics and Supplementary Tables) or if technical issues (i.e., imaging and/or electrophysiology artifacts) occluded quantification of the SD (described in the Electrophysiology and Imaging sections, below).

Figure 1. Regional heterogeneity of cerebral perfusion responses to SD.

A. Experimental timeline and design for study. Before dMCAo or sham surgeries, burr holes were prepared for electrophysiology and SD inductions (arrowheads; ~20 min. between each SD). Imaging (LSCI and GCaMP5G fluorescence) and local field potential (LFP) recording sessions began shortly after distal middle cerebral artery occlusion (dMCAo) or sham surgeries and continued throughout the experiment (solid and dotted lines). Filled symbols on timeline indicate the time points of animal exclusions during LSCI (blue, n = 4 mice) and GCaMP5G (green, n = 6 mice) experiments. Text box on right shows mouse numbers that were included for each data set shown in Figures. ⁺Poor imaging quality (n = 2 mice) prevented accurate analysis GCaMP5G imaging during all three SDs and were thus excluded from Ca²⁺ (Figure 3C), but not DC shift, data sets (Figure 4B). B. Diagram showing regions of interest (ROIs) used for LSCI (and GCaMP5G, in Fig.3) analysis labeled “remote” and “proximal” with respect to their relative proximity to the dMCA occlusion site (inset image). C. Representative pseudo-colored LSCI maps (images) and group data (normalized to the contralateral hemisphere) show decreased ipsilateral perfusion 50 min. after dMCAo (n = 7 mice shown) compared to sham (n = 7 mice). Ipsilateral perfusion was similar to the contralateral hemisphere and did not differ between the two ROI locations in sham mice (P > 0.9). After dMCAo, perfusion deficits were seen in both ROI locations (vs. sham) and were further augmented in proximal locations (remote vs. proximal in dMCAo). The perfusion deficit seen in one animal from the dMCAo group (filled data point at 0.625, remote) was an outlier and excluded from statistical analysis (see Supplemental Table 1B). D. Traces show representative perfusion responses in remote and proximal ROIs during the first SD (~60 min. post dMCAo) from an individual animal (arrowheads indicate the onset of the SD-induced cerebral blood flow response). Group data from the same animals shown in C quantifies the magnitude of hypoperfusion ((a) in traces) and hyperperfusion ((b) in traces) components during SD. Hypoperfusion was similar across ROIs in sham and dMCAo experiments while hyperperfusion after SD was significantly blunted only in proximal ROIs from dMCAo mice (n = 7 SDs per ROI location from n = 7 mice). See Supplementary Table 1 for all statistics related to Figure 1. * P<0.05, ** P<0.01, ***P<0.001, **** P<0.0001 from Bonferroni’s multiple comparisons test.

Surgical preparation and distal middle cerebral artery occlusion

Urethane (1.5 mg g⁻¹ induction dose administered i.p. with ≤ 0.75 mg g⁻¹ for maintenance, as needed) was chosen for anesthesia as isoflurane has central nervous system (CNS) -depressive effects (Peterson, Drummond et al. 1986) and itself suppresses SD (Kudo, Toyama et al. 2013). Urethane-anesthetized mice undergoing prolonged in vivo experiments have increased mortality rates when combined with airway obstruction by stereotaxic frames (Moldestad, Karlsen et al. 2009). Therefore, we provided additional supplementation with compressed room air via the stereotaxic frame gas inlet port throughout recording sessions as an alternative to additional invasive procedures (i.e., tracheotomy). After anesthesia induction (loss of toe-pinch reflex), mice were placed in a small animal stereotaxic frame assembly (David Kopf Instruments, Tujunga, CA, USA), and body temperatures were maintained at 36.5-37°C using a feedback-controlled heating pad system (Kent Scientific Corporation, Torrington, CT, USA). After head fixation, sterile eye ointment (PuraLube Vet Ointment, Dechra Veterinary Products) was applied and the skull was exposed and kept moist with saline (0.9% NaCl) and mineral oil-soaked cotton to improve image quality (Chung, Sugimoto et al. 2018). Burr holes were drilled (1 mm diameter) and a small portion of dura was carefully removed by microdissection for SD induction (ML: + 1.0 mm, AP: + 3.0 mm, relative to bregma) as well as for placement of recording electrodes (in two sites within the MCA territory AP: − 2.0 mm, ML: 1.0 and 2.0 mm, relative to bregma) in the right hemisphere.

After burr hole preparation, focal stroke was induced by occluding the right distal middle cerebral artery (dMCAo) via cauterization, as described previously (Kuraoka, Furuta et al. 2009, Lindquist and Shuttleworth 2014). Briefly, an incision in the skin was made between the lateral part of the orbit and external auditory meatus, and the temporalis muscle was removed. A small craniotomy (~2 mm) was drilled on the frontal bone just rostral to the zygomatic and squamosal bone union to allow transtemporal exposure of the MCA for coagulation (see Figure 1B). In sham-operated animals, the MCA branch was exposed but not cauterized. After occlusion or sham procedures, animals stabilized for 60 min before SD challenge. During this time recording electrodes were gently positioned into burr holes in the cortex (see Electrophysiology, below) and sterile Ringer’s lactate solution was administered (5ml/kg, s.c.) to minimize fluid loss throughout the experiment (each lasting ~3.5 – 4.5 hours). After the experiment, the animal was euthanized by pentobarbital overdose (i.p. administration of 0.02cc, 260 mg ml⁻¹; Fort Dodge Animal Health, Fort Dodge, IA) without awakening from anesthesia.

SD generation

The first SD likely occurred shortly after the onset of occlusion in all animals (Shin, Dunn et al. 2006), but went undetected in our experiments as this was before the establishment of electrode and imaging recordings. Animals were allowed to recover for one hour from the initial occlusion or sham procedure. SDs were then initiated by microinjection (~1μl) of 1M potassium chloride (KCl) to provide reproducible sets of SDs that, in the case of dMCAo mice, propagated through the infarcted hemisphere with regions of different metabolic capacity. KCl-evoked SDs were generated as previously described (Aiba, Carlson et al. 2012) at a rostral location, distant from the MCA territory (see Surgical Preparation and Figure 1B). After visual confirmation of each SD (using imaging techniques, described below), any residual KCl was quickly removed (absorbed with a cotton-tip applicator) to prevent subsequent SD initiation. The first induced SD was evoked ~60 minutes following dMCAo or sham surgeries and there were no significant differences in the onset time of KCl-induced SDs between groups (Figures 1 & 2; sham: 64.4 ± 2.0 min., dMCAo: 66.5 ± 3.1 min.; one-way ANOVA P = 0.6; n =7 mice/group. Figures 3 & 4; sham: 61.9 ± 0.9 min (n = 3 mice); dMCAo + vehicle: 61.9 ± 3.4 min (n = 6 mice); dMCAo + ketamine: 61.5 ± 1.4 min (n = 7 mice); P > 0.9). We examined tissue consequences during repetitive SDs (a total of 3) separated by 15-20 minute intervals (Figure 1A), as SDs that occur in clusters are particularly challenging for the injured brain (Dreier, Fabricius et al. 2017, Hartings, Andaluz et al. 2020). Under these conditions, spontaneous SDs were infrequently recorded (n = 1 animal from the present study), likely because clusters of KCl-induced SDs were generated before the establishment of spontaneous events (von Bornstadt, Houben et al. 2015). The experimental strategy employed here was designed similarly to early work by Busch et al., which demonstrated step-wise lesion expansion using MRI with each SD evoked at 15 minute intervals after MCA occlusion (Busch, Gyngell et al. 1996), see Discussion.

Figure 2. Prolonged DC shift durations in proximal brain regions after dMCAo.

A. Schematic showing the relationship of LSCI ROIs (black and red circles, analysis in Figure 1) with respect to remote and proximal LFP recording locations and SD induction site. B. Traces show DC potential shifts during repetitive SDs (arrowheads) in remote (black) and proximal (red) recording locations after dMCAo and sham procedures (double lines indicate each SD was separated by 15 - 20 min). C. Data from the same animals shown in Figure 1, demonstrate that DC shift durations were prolonged in proximal locations after dMCAo. In one sham animal, the final SD stimulation took longer than 20 min. and was excluded from analysis. In one animal from the dMCAo group, the DC shift during the 1^st SD (proximal) was prolonged (data point at 920s) and DC shifts for subsequent stimulations were not detectable in this same location thus preventing quantification of duration. In this same animal, the 3^rd SD in the remote location was also not detected. In a second dMCAo animal with an extended proximal DC shift (data point at 880s), the 3^rd SD stimulation also had no detectable DC shift. In total, plot shows n = 20 SDs each in proximal and remote locations from n = 7 sham mice and shows n = 20 SDs (remote) and n = 18 SDs (proximal) from n = 7 dMCAo mice. DC shift duration data followed a lognormal distribution and thus Y = log(Y) was used to transform the data for use in parametric tests. D. DC duration versus the maximum hyperperfusion response (from the same animals in Figure 1D(b)) shows that prolonged DC shifts in proximal locations were also characterized by blunted hyperperfusion during SD. Symbols plotted >500s were outliers also shown in C. *** P < 0.001, **** P<0.0001, n.s. denotes P > 0.05, from Bonferroni’s multiple comparisons test (see Supplementary Table 2).

Figure 3. Ketamine reduces intracellular neuronal Ca²⁺ load during SD.

A. (Right) average projection of 10 frames taken ~50 minutes after dMCAo in a vehicle-treated GCaMP5G-expressing animal. Image shows the location of recording electrodes, Ca²⁺ ROIs, and SD initiation site relative to bregma (b) and the dMCA. B. Top: Ca²⁺ traces during repetitive SDs (arrowheads) in a vehicle-treated dMCAo animal show an example of prolonged elevations (red trace) in neuronal Ca²⁺ during the first SD and failure of subsequent SDs to propagate in the same location. In this animal, neurons in remote (black trace) brain regions had progressively worse recovery of SD-induced Ca²⁺ elevations during the SD cluster. SDs in the presence of ketamine showed delayed progression of irrecoverable Ca²⁺ transients in proximal ROIs with no change in remote locations during repetitive SDs. Fluorescence changes are normalized to levels prior to SD1 (dotted lines). C. Group data showing Ca²⁺ signal half-life (t₅₀) during repetitive SDs. In this data set, two animals (one each from vehicle- and ketamine- treated dMCAo groups), were excluded due to poor imaging quality which prevented accurate Ca²⁺ analysis. In one animal from the dMCAo + vehicle group, Ca²⁺ increases remained elevated from the 2^nd SD preventing quantification of the 3^rd SD (proximal location). 1 SD (filled symbol, remote location) was an outlier. Plot shows n=17,14 SDs from remote and proximal ROIs, respectively for the dMCAo + vehicle group (n = 6 mice). In one animal from the dMCAo + ketamine group (n = 6 mice), the 3^rd SD stimulation did not induce Ca²⁺ changes in the proximal ROI. Figure shows data from n = 18,17 SDs from remote and proximal ROIs, respectively. In one animal from the sham group, imaging artifacts emerged after the 1^st SD stimulation that prevented the measurement of subsequent SDs (both ROIs). Data shows n = 7 SDs in remote and proximal locations from n = 3 mice. *P <0.05, **** P < 0.0001. See Supplementary table for full statistical reports.

Figure 4. Ketamine can decrease the duration of depolarization during SD.

A. DC shift traces during repetitive SDs in a ketamine-treated animal shows that ketamine can reduce the duration of SD in proximal recording locations after dMCAo. B. Group DC shift durations from the same animals shown in Figure 4 are shorter in ketamine-treated animals after dMCAo. Sham: n = 9 SDs (proximal and remote) from n = 3 animals, dMCAo + vehicle: n = 15 SDs (remote) and n = 15 SDs (proximal) from n = 5 animals, dMCAo + ketamine: n = 21, 20 SDs (remote and proximal, respectively) from n = 7 animals. Exclusions: one animal from the dMCAo + vehicle group had a spontaneous SD, and the DC potential did not recover to >80% of baseline prior to subsequent evoked SD stimulations (both recording locations, data not shown); from the dMCAo + ketamine group the DC potential during one SD did not recover to >80% prior to the next SD stimulation (data not shown). Group comparisons were conducted on transformed DC shift data (as in Figure 2), see Supplementary Table 4 for full statistical reports. C. DC shift duration versus Ca²⁺ half-life during SD demonstrates that slower recovery of Ca²⁺ is related to the duration of depolarization within the same animal. Note the different x and y-axis scales between graphs: the dotted box in dMCAo ketamine (middle) shows the relationship to sham animals (left) while the two dotted boxes in the dMCAo + vehicle group show both sham and ketamine-treated dMCAo scales. Data is from the same animals in B that had both calcium and DC shift data. *P<0.05, ***P<0.001.

Imaging

Laser speckle contrast imaging (LSCI) acquisition began shortly after (~10 minutes) dMCAo or sham surgeries and was used to monitor blood flow changes during SD, as previously described (Lindquist and Shuttleworth 2014). Briefly, the intact exposed skull was illuminated by a 785 nm laser diode (Thorlabs, Newton, NJ, USA), and reflected backscattered light passed through a long pass-filtered (720 nm) and focused with an SLR camera lens (55 mm; Nikon, Tokyo, Japan). Images were collected at ~ 6 Hz using a digital CCD camera (Stingray F-504B, Allied Vision Technologies, Stadtroda, Germany) and cerebral perfusion maps were calculated and displayed in real-time by LabVIEW software modified from Bernard Choi, UCI (Yang, Cuccia et al. 2011). Offline calculations in Fiji (RRID:SCR_002285) (Schindelin, Arganda-Carreras et al. 2012) were used to ratio blood flow in ipsilateral regions of interest (ROIs) to corresponding ROIs on the contralateral hemisphere. ROIs were ~750μm in diameter (avoiding major blood vessels) and created in cortical areas close to (“proximal”) and more distant from (“remote”) the ischemic MCA territory (see schematic in Figure 1B). ROI coordinates were similar in sham and dMCAo groups for remote (ML: 1.4 ± 0.05 & AP: −1.3 ± 0.09 mm in sham versus ML: 1.5 ± 0.06 & AP: − 1.3 ± 0.08 mm in stroke, P > 0.9 for both ML and AP, relative to bregma, n = 7 animals/group) and proximal locations (ML: 3.5 ± 0.04 & AP: −1.3 ± 0.08 mm in sham versus ML: 3.5 ± 0.06 & AP: −1.2 ± 0.07 mm in stroke, P > 0.9 for both ML and AP, relative to bregma, n = 7 animals/group). LSCI experiments were conducted in a total of 18 mice, n = 4 mice were excluded (n = 3 dMCAo mice did not reach primary endpoint, n = 1 sham surgical error, see Figure 1A). Results show LSCI data from 8 males, 6 females (4 WT and 3 GCaMP5G mice in each group) totaling 7 mice/group (Figure 1).

A custom LED-based system was built to monitor wide-field fluorescence through the intact skull GCaMP5G-expressing mice. LEDs and associated optical components were purchased from Thorlabs (Newton, NJ, USA). A 470 nm LED was collimated and mounted to a cage cube containing a 505 nm dichroic long pass (DCLP) beam splitter and incident light was orthogonal to the dorsal surface of the exposed skull. Fluorescent signals were long pass (505 nm DCLP) and bandpass filtered (510/40nm). Images (1280 x 960 pix, 1 x 1 binning) were collected at 4 Hz using a CCD Mightex camera controlled by Micro-Manager open-source microscopy software (RRID:SCR_000415) (Edelstein, Amodaj et al. 2010). ROIs (500 μm diameter) on dMCAo or sham ipsilateral (right) hemispheres were normalized to corresponding contralateral ROIs and values were then transformed to reflect the change in fluorescence from baseline (ΔF/F₀; F₀: average intensity of 10 baseline frames prior to each SD). GCaMP5G signals during SD had contaminating signals ~30s after the peak and during the second phase of the transient. These are likely artifacts due to blood flow changes and/or intrinsic optical signals arising from parenchyma tissue undergoing SD (Zhao, Tuohy et al. 2021). To exclude these contributions from our GCaMP5G analyses and examine neuronal kinetics of recovery from SD, the peak amplitude of transients were aligned, and curves were fit using nonlinear regression algorithms (from the peak to 30s after peak). From these curves, recovery of fluorescence to 50% of mean baseline signals was calculated and reported as the half-life of recovery (t₅₀). Individual SDs were excluded if fluorescence did not increase ≥ 5 standard deviations above pre-SD baseline ΔF/F₀ values or if transients did not recover to 50% (of pre-SD values). A total of 17 GCaMP5G-expressing mice (9 males, 8 females) were utilized for Ca²⁺ imaging experiments. Poor imaging quality (n = 2 animals, one each from vehicle- and ketamine- treated dMCAo groups) prevented accurate analysis of all three SDs and were thus excluded from Ca²⁺ (Figure 3C), but not DC shift, data sets (Figure 4B). Figure 3 shows data from n = 6 dMCAo + vehicle, n = 6 dMCAo + ketamine, and n = 3 sham + vehicle. In this data set n = 6 mice did not reach the primary endpoint or inclusion criteria and were therefore excluded (Figure 1A).

Electrophysiology

Glass micropipettes containing Ag/Cl wires were filled with artificial cerebral spinal fluid (aCSF; containing (in mM): NaCl, 126; NaHCO₃, 26; glucose, 10; KCl, 3; CaCl₂, 2, NaH₂PO₄, 1.5; MgSO₄, 1) or normal saline (0.9% NaCl) and carefully placed into the two burr holes within the MCA territory (see surgical preparation, above) at a depth of ~500μm (from the skull surface) for continuous recording of local field potentials (LFP). Recordings were referenced to a Ag/Cl ground wire placed under the skin through a small incision at the back of the neck. Signals were amplified using an Axoclamp-2B amplifier (Molecular Devices, Sunnyvale, CA, USA) and 1X gain outputs were A/D- converted and displayed using a PowerLab 8/35 digitizer and LabChart 7 software, respectively (AD Instruments, Dunedin, New Zealand). A lowpass filter (0.5 Hz) was used for DC recordings and DC potential shift durations were calculated from onset (drop below baseline voltage) to 80% recovery. Individual SDs were excluded if the DC potential did not drop below 20% of baseline voltage or recover to 80% before the next SD stimulation. Results in Figure 2 show DC shift durations recorded from animals used also in LSCI experiments (n = 7 mice/group). Adequate electrophysiological recordings for DC shift analyses (Figure 4B) were collected in 3/3 sham, 5/7 dMCAo + vehicle, and 7/7 dMCAo + ketamine mice from GCaMP5G imaging experiments.

Drugs

Urethane was purchased from Sigma-Aldrich (PubChem Substance ID: 24900632; Merck KGaA, Darmstadt, Germany) and dissolved daily in normal saline (0.9% NaCl). Ketamine (100 mg/ml, racemic: R (−)/S (+), Putney, Inc., Portland, ME), was diluted in saline and administered (i.p.) at a dose of 5 mg kg⁻¹.

Statistical analysis

Power analyses were conducted in G*Power 3.1 (RRID:SCR_013726) (Faul, Erdfelder et al. 2009). The sample size for examining the effects of ketamine on DC shift duration was determined from posthoc calculation of achieved power from our published data showing that 30μM ketamine reduced DC shift durations without blocking SD in acute brain slices (Reinhart and Shuttleworth 2018). From data shown in Supplementary Figure 1, control DC shift durations (SD2) were 38.7 ± 8.3s (mean ± SD, n=8 slices from 6 animals, Fig 1B) and DC shift durations in ketamine were 27.1± 5.5s (mean ± SD, n=6 slices from 5 animals, Fig 1A). Using the t-test family comparing two independent means and SD = 7, the calculated effect size was 1.66. The achieved power in this data set (two-tailed, d = 1.66, α = 0.05) was 0.81. Using an effect size of 1.66 (two-tailed, α = 0.05, N2/N1 = 1) n = 7 samples/group are required to achieve >80% power.

The sample size for evaluating the effects of ketamine on neuronal Ca²⁺ loading was determined from data in Figure 5B (Reinhart and Shuttleworth 2018) where ketamine (30μM) reduced neuronal Ca²⁺ accumulation during SD (measured from the integral of GCaMP5G fluorescence increases) in metabolically compromised (vulnerable) brain slices. Using the F-test family (analysis of variance (ANOVA), fixed effects, omnibus, one-way) comparing control (5057±1288 ΔF/F₀ x Δt, n=8 slices), vulnerable (7268±917.3 ΔF/F₀ x Δt, n=6), and vulnerable + ketamine (5312±1429 ΔF/F₀ x Δt, n=6) Ca²⁺ accumulation measurements (mean and SD = 1200) the calculated effect size was 0.81. The achieved power in this data set (effect size = 0.81, α = 0.05) was 0.87. Using an effect size of 0.81(α = 0.05) n = 7 mice/group are required to achieve >80% power (actual power = 0.87).

Data are reported as mean ± SEM. Statistics were performed in GraphPad Prism (RRID:SCR_002798; version 9, San Diego, CA, USA). All outlier, normality and lognormality, and parametric test results (including F values, degrees of freedom (DF), and P values) are provided in Supplementary Tables according to each figure in the Results. The ROUT method for the calculation of outliers (Q=1%) was used before all statistical tests. Outliers are reported in Figure Legends and in Supplementary Tables and were not included for group comparisons of statistical significance. D’Agostino & Pearson Test was utilized as the primary test for normality and lognormality; however, for data with lower sample sizes (i.e., as in Figure 1C, Figure 1D, and sham groups in Figure 3) the Shapiro-Wilk test was employed. In DC shift duration data sets, multiple groups violated the normality assumption and instead were likely sampled from a lognormal distribution (see Supplementary Tables 2 & 4). DC shift durations are plotted (in seconds) in Figure 2C and Figure 4B; however, data were transformed (using Y=log(Y)) and the transformed values were used in parametric tests. Parametric tests evaluating statistical significance included unpaired t-tests and one- and two-way ANOVA with Bonferroni’s correction during multiple comparisons. P values < 0.05 were considered significant.

RESULTS

Regional heterogeneity of cerebral perfusion responses to SD after dMCAo

After stroke, brain tissue surrounding the infarcted region is often viable but at risk for succumbing to SD-mediated injury. To better understand mechanisms that lead to injury progression in vulnerable tissue during SD we performed dMCAo in mice. In this model, transtemporal access of a distal branch of the MCA allowed for occlusion under direct visualization (representative image of MCA shown in Figure 1B) and resulted in perfusion gradients across cortical areas supplied by the MCA watershed; including a necrotic core, penumbra, and intact/undamaged brain regions (Kuraoka, Furuta et al. 2009). Representative images created using LSCI are displayed in Figure 1C and show blood flow maps normalized to the contralateral hemisphere at approximately 50-60 minutes post sham and dMCAo surgeries. Group data in Figure 1C show baseline cerebral perfusion deficits after dMCAo are moderately exacerbated in ROIs proximal to the infarction compared to more remote ROIs (see Supplementary Table 1 for Figure 1 statistics). These data confirm regional differences in metabolic capacity across the cortex. ROIs even closer to the site of MCA infarction, but still within the imaging field on the surface of the cortex, had a further reduction in cerebral perfusion (~32 ± 1.6% decrease from the contralateral hemisphere, n = 7 mice) compared to remote and proximal ROIs (~16 ± 0.9% and ~27 ± 2.0% reduction for remote (n = 6) and proximal (n = 7), respectively). This was a significant decrease in cerebral perfusion compared to remote (P < 0.0001), but not proximal, ROI locations (P = 0.13) in dMCAo mice (one-way ANOVA, DF = 17, F= 24.98, P < 0.0001).

To evaluate the hemodynamic responses to SDs, we generated repetitive SDs at ~15 - 20 min intervals (19.5 ± 1.7 and 17.1 ± 0.7 min intervals for sham and dMCAo groups, respectively; unpaired t-test (two-tailed): n = 13 SDs, DF = 24, F = 6.239, P = 0.1988). Focal microinjections of KCl into a rostral burr hole distant from the immediate MCA territory were used to initiate SDs (see Methods; Figure 1B). The vascular response to SD is complex and varies from species to species (see review (Ayata and Lauritzen 2015)). In line with previous work (Ayata and Lauritzen 2015), the first SD evoked in sham animals was characterized by a pronounced hypoperfusion/vasoconstriction (coinciding with the DC shift) followed by a small and transient hyperperfusion/dilation and more long-lasting oligemia (Figure S1A). Subsequent SDs in these mice (Figure S1) were then largely distinguished by pronounced hyperperfusion with minimal to no preceding hypoperfusion (Ayata and Lauritzen 2015). In dMCAo mice, relative perfusion changes during the first evoked SD (initiated 60 min. post-surgery) were characterized by hypo- with minimal hyper-perfusion in proximal ROIs, compared to slightly more biphasic transients in less ischemic (remote) ROIs in dMCAo mice (traces in Figure 1D). Results in Figure 1D(b) show that SD-induced hyperperfusion responses during the first SD were reliably blunted only in proximal locations in dMCAo animals (n = 7 SDs per ROI location from 7 mice, Two-way ANOVA, Supplementary Table 1). The magnitude of hypoperfusion responses to SD was larger in proximal locations than in remote areas from both sham and dMCAo animals, indicating a smaller impact of ischemia on this parameter (Figure 1D(a), n = 7 SDs per ROI location from 7 mice. Two-way ANOVA, Supplementary Table 1).

More remote regions after stroke were initially no different than respective remote ROIs in sham animals (Figure 1D). However, hyperemic responses became less prominent during repetitive SD events (traces in Figure S1A and group data in Figure S1B(b)). Only dMCAo animals showed less pronounced perfusion responses (both hypo- and hyper-perfusion) during repetitive SDs in proximal ROIs (Figure S1). This could be due to lesion expansion or other mechanisms that inhibit SD propagation in these regions. In sham animals, successive SDs (SDs 2 and 3; Figure S1) were initiated on an already oligemic background (resulting from the first SD stimulation, as reviewed by (Ayata and Lauritzen 2015). This suggests that cerebral blood flow (CBF) measurements of hypoperfusion during SD may be less diagnostic for determining the metabolic capacity of underlying tissue as it could reflect either ischemic injury progression or normal vascular responses during SD clusters. Collectively, these data show that brain tissues proximal to the MCA occlusion are less able to mount hyperemic responses after SD – which indicates impaired neurovascular coupling and reflects vulnerability to further injury in these regions (Feuerstein, Takagaki et al. 2014).

DC shift durations are prolonged in proximal regions

SDs characterized by prolonged durations of depolarization can be indicative of underlying tissue being metabolically compromised (Lindquist and Shuttleworth 2014, Dreier, Fabricius et al. 2017, Hartings, Shuttleworth et al. 2017). We thus monitored DC shift durations during repetitive SDs using electrodes placed in remote and proximal ROIs (diagram in Figure 2A). Representative DC shifts during SD in dMCAo and sham mice are shown in Figure 2B and group data in Figure 2C (see statistics in Supplementary Table 2). These data show that DC shift durations are significantly longer in proximal recording locations after dMCAo (n = 20 SDs in remote locations in both sham and dMCAo; n = 20, 18 SDs in proximal locations from sham and dMCAo, respectively). In two mice from the dMCAo group, the DC potential shift was significantly prolonged in proximal tissues (data points > 400s in Figure 2C) and DC shifts were less distinguishable during subsequent SDs. In total, ~86% of the KCl-induced SDs successfully propagated (confirmed via LSCI and/or DC recordings) into proximal tissues (18 confirmed SDs / 21 stimulations) compared to remote regions where virtually all SDs propagated through (20 confirmed SDs / 21 stimulations; n = 7 mice) in dMCAo animals. In sham experiments, 100% of KCl evoked SDs were detected in both regions (20/20 remote and proximal from n = 7 mice; the final SD stimulation in one animal was generated >30 min. after the previous SD and was thus excluded from both remote and proximal locations for the 3^rd SD, Figure 2). Comparisons of hyperemic responses (shown in Figure 1) and respective DC shift durations of the SDs are plotted in Figure 2D and demonstrate that prolonged depolarizations are accompanied by reduced tissue ability to increase blood flow (i.e., supply-demand mismatch).

Ketamine reduces neuronal Ca²⁺ and DC shift durations after dMCAo

In brain slice models of metabolic compromise, SD induces irrecoverable neuronal Ca²⁺ loading that can be reduced by NMDAR antagonists, including ketamine, to improve neuronal recovery (Aiba and Shuttleworth 2012, Reinhart and Shuttleworth 2018). To investigate this in vivo, we monitored cortical neuron Ca²⁺ accumulation during SD through the intact skull of GCaMP5G-expressing mice (Figure 3). Ketamine (5 mg kg⁻¹) or saline vehicle control was administered (i.p.) 30−45 minutes after dMCAo or sham surgery based on reports showing maximum plasma concentrations of ketamine are achieved within ~30 minutes after injection (i.p.) (Toki, Ichikawa et al. 2018) and closely resemble cerebral spinal fluid (CSF) concentrations (Khlestova, Johnson et al. 2016, Toki, Ichikawa et al. 2018). This dose was also chosen based on pilot experiments (n = 3 mice, data not shown) showing this dose did not block SD initiation. In these experiments, SD intervals were again not different between groups. In vehicle-treated dMCAo mice (n = 6 mice), ~78% of SD initiation attempts (14/18) resulted in Ca²⁺ transients observed in proximal ROIs compared to 94% (17/18) within remote regions. Traces in Figure 3B display an example of irrecoverable Ca²⁺ signals during the first SD that become lost during subsequent SDs. Additionally, remote ROIs show step-wise increases in the duration of neuronal Ca²⁺ that resemble more proximal areas by the final SD in the cluster. In ketamine-treated animals, SD-induced Ca²⁺ signals were confirmed in both remote and proximal ROIs during most SD attempts (18/18 for remote, 17/18 proximal from n = 6 mice). Results in Figure 3C show the time course of recovery was accelerated in proximal penumbral regions in the presence of ketamine compared to delayed recoveries in proximal locations in vehicle-treated mice (Two-way ANOVA, see Supplementary Table 3 for statistics). However, the Ca²⁺ load in ketamine-treated animals remained elevated compared to proximal regions in non-ischemic animals. Traces in Figure 3B show that Ca²⁺ dysregulation is still observed in proximal locations after ketamine treatment likely due to the immense challenge that SD clusters pose to vulnerable brain. However, ketamine did prevent the progression of irrecoverable Ca²⁺ accumulation in remote regions during clustered SDs (Figure 3B traces). DC recordings during SD showed that ketamine treatment could reduce the duration of depolarization in proximal locations (Figure 4) during repetitive SDs (Two-way ANOVA, Supplementary Table 4). Comparisons of Ca²⁺ recovery half-life to the duration of DC shifts indicated that ketamine improved potentially damaging consequences of neuronal Ca²⁺ loading in parallel with reducing the DC shift duration after dMCAo.

DISCUSSION

General Findings

The main finding of this study is that ketamine can substantially improve the recovery of brain tissue after the metabolic challenge of SD, in a mouse model of focal stroke. This study involved the experimental generation of SD at a remote site and evaluated the consequences of SD as it propagated through a gradient of metabolic capacity, defined by distance from the core of the ischemic lesion. Together with dual-site microelectrode recordings, this design afforded the opportunity to relate the duration of neuronal depolarization, Ca²⁺ loading, and hemodynamic responses to SD at different sites relative to the perfusion deficit. The results are consistent with the conclusion that longer depolarizations and increased Ca²⁺ loading are associated with impaired neurovascular responses in locations with reduced baseline perfusion. At a concentration of ketamine that did not impair the initiation or propagation of SD, the antagonist effectively reduced the duration of SDs and accelerated the recovery of neuronal Ca²⁺ loading. These in vivo findings suggest that protective effects of ketamine against SD-mediators of injury could be achieved without necessitating the use of high doses.

Ketamine to target SDs in injured brain

Ketamine has emerged as a leading candidate for the clinical treatment of SD in patients with acute brain injury. Initial studies noted that switching to ketamine as a sedative agent was associated with a decrease in the number of SD events in ICU patients (Sakowitz, Kiening et al. 2009), a finding confirmed and extended in a subsequent retrospective analysis (Hertle, Dreier et al. 2012). A prospective study then demonstrated the effectiveness of ketamine in reducing SD incidence in brain-injured patients (Carlson, Abbas et al. 2018) and a multi-site trial is currently underway testing outcomes following the reduction of SDs by ketamine (NCT05337618). These clinical efforts are based on very strong evidence that suggests that inhibiting invasion of SD through vulnerable tissue will reduce neuronal cell death and reduce infarct expansion. However, and as noted above (Introduction), high ketamine concentrations required to block SD also cause marked sedation which is associated with increased risk of ICU complications (Abou-Chebl, Lin et al. 2010).

In rodents, intraperitoneal injection of ketamine up to 30 mg kg⁻¹ is considered subanesthetic and is used in animal models examining ketamine’s rapid anti-depressive effects (Zanos, Moaddel et al. 2016, Zanos and Gould 2018). Recent work has shown that up to 100mg kg⁻¹ (i.p.) still does not induce marked sedation in mice, and is utilized for studying the ‘dissociative’ effects of ketamine (Zanos and Gould 2018, Cichon, Wasilczuk et al. 2023). Additionally, the potency of ketamine blockade of NMDARs depends not only on receptor subunit composition but also on ketamine stereospecificity (Khlestova, Johnson et al. 2016, Zanos and Gould 2018). Both stereoisomers of ketamine S (+) and R (−) are clinically used; however, the S (+) isomer is ~ 2 times more potent at NMDARs (Peltoniemi, Hagelberg et al. 2016). Here we used a 5 mg kg⁻¹ dose of racemic ketamine which is well-below anesthetic doses in rodents, and is also expected to correspond to subanesthetic doses in humans (~0.5 mg kg⁻¹; (Khlestova, Johnson et al. 2016)).

A striking finding of the current study was that the low dose of ketamine tested here could be very effective at reducing the duration of DC shifts in the locations close to the ischemic core (“proximal” regions in Figure 4). In contrast to the relatively limited effectiveness of NMDAR block of SD initiation/propagation in ischemic tissues or models of metabolic impairment (e.g., hypoxia or oxygen-glucose deprivation models), the inhibition of the NMDAR-dependent late phase of SD has the potential to be very effective at protecting neurons in these more ischemic locations. This finding suggests promise for clinical targeting of SD consequences in ischemic brain tissues.

Ca²⁺ dynamics

Ketamine reduced intracellular Ca²⁺ loading in neurons and the duration of depolarization during SD (Figures 3&4). We detected DC shifts and/or Ca²⁺ transients in nearly all animals undergoing repetitive SD challenge (in sham and ketamine-treated dMCAo groups) suggesting that underlying tissue is viable as it can sustain subsequent SDs (Koroleva and Bures 1996). Clusters of SDs are particularly challenging for compromised tissues and it is thus not surprising that irrecoverable Ca²⁺ transients were detected in a few animals, despite the beneficial effects of ketamine. Nevertheless, ketamine exposure did delay the progression of injury into remote regions (Figure 3). These data are consistent with our prior findings using ketamine in metabolically vulnerable brain slices with SDs (initiated by focal K⁺); where neuronal Ca²⁺ influx was attenuated by ketamine (30μM) without preventing SD propagation (Reinhart and Shuttleworth 2018). In those studies, ketamine improved recovery of evoked neurotransmission and brain slices were capable of propagating subsequent SD waves, indicating the presence of viable neurons (Koroleva and Bures 1996).

Metabolic heterogeneity in response to SD

Using LSCI to monitor CBF throughout the entire cortex, we confirmed regional differences in perfusion (prior to SD stimulations) in dMCAo animals (Figure 1). Due to technical limitations (i.e., repositioning of animals for imaging and electrophysiology after surgery) we did not collect baseline perfusion maps prior to dMCAo or sham surgeries. Therefore, residual blood flow compared to baseline levels was not used to determine stroke core and penumbra areas based on previous literature. Instead, ROIs were normalized to their respective locations in the intact contralateral hemisphere for analyses, and cortical areas with mild (remote) and moderate (proximal) levels of perfusion deficits were identified. ROIs placed as close as possible to the site of surgical occlusion and still within the imaging field (~ 4-5mm lateral to bregma on the skull surface) showed a slight but not significant decrement in perfusion compared to proximal ROIs (i.e., ~32% vs. ~27%). This could indicate that proximal ROIs were indeed located in viable penumbral areas and that more severe perfusion deficits exist outside of our field of view and represent the true infarct core.

In the naïve otherwise healthy cortex in mice, intact neurovascular coupling in response to a single SD is most often characterized by profound initial hypoperfusion which is followed by transient hyperperfusion and long-lasting oligemia (as shown in Figure S1 traces during the first SD in a sham animal). On this oligemic background, we saw that subsequent SDs did not further reduce cerebral blood flow, but did have a hemodynamic response that was predominately hyperemic, as previously described (Ayata and Lauritzen 2015). Though dMCAo did not drastically alter the magnitude of hypoperfusion during SD, we did observe a marked reduction in the degree of hyperperfusion responses to SD in proximal ROIs (Figure 1D) that was consistent during repetitive SDs (Figure S1). Initially, remote regions after dMCAo were capable of generating hyperperfusion responses to SD, that were similar in magnitude to those seen in sham animals (i.e., no differences seen in Figure 1D(b)). However, there was a progressive loss of SD-induced hyperperfusion in remote regions in dMCAo animals during repetitive SDs, which was not observed in sham animals (Figure S1). This transition of SD hemodynamics relates to tissue metabolic capacity and has been observed after ischemia in animals (Strong, Bezzina et al. 2006, Feuerstein, Takagaki et al. 2014, Ayata and Lauritzen 2015) and in humans after malignant middle cerebral artery infarction (Woitzik, Hecht et al. 2013). Differences in the hemodynamic responses may be attributed to pathological mechanisms preventing hyperemia (Ayata and Lauritzen 2015).

A recent study has described neurovascular dynamics occurring during SDs in a photothrombotic stroke model and show a clear worsening of SD-evoked vasoconstriction in more ischemic regions during repetitive SDs (Zhou et al., 2021). These findings also have implications for the mechanism of vasogenic edema development after stroke, which can lead to life-threatening swelling in large strokes. Recent preclinical work has demonstrated that the source of this vasogenic edema is related to CSF influx into the periarteriolar space that is created due to vascular collapse during spreading ischemia (Mestre, Du et al. 2020). It is therefore plausible that reversal of this vascular collapse with ketamine, even without complete SD suppression could decrease such malignant edema formation. SD also appears to contribute to infarction even when perfusion is restored after stroke (Torteli et al., 2023), and it will be of interest to determine whether reducing the duration of SDs with ketamine improves reperfusion.

Conclusions

In this study, we saw increased vulnerability of neurons during SD in vivo that was based on proximity to the infarction. In proximal regions, insufficient vascular supply and prolonged DC shift durations can contribute to severely impaired neuronal recovery following SD. Accumulation of intracellular neuronal Ca²⁺ was enhanced in vulnerable brain regions and serves as potential cellular targets for limiting excitotoxicity during SD. This is supported by our findings that the NMDA receptor antagonist ketamine could reduce the duration of DC shifts and neuronal Ca²⁺ loading, and improve neuronal recovery following SD.

Supplementary Material

SUPINFO

Supplementary Figure 1. A. Cerebral perfusion responses during repetitive SDs. Traces from LSCI experiments showing representative perfusion responses during repetitive SD stimulations (arrowheads) in sham and dMCAo animals. B. Group data from animals shown in Figure 1 (n = 7 mice). Data in (a) demonstrates no significant change in hypoperfusion responses during successive SDs in remote locations (n = 20, 19 SDs for sham and dMCAo, respectively). However, in proximal locations (n = 17, 19 SDs for sham and dMCAo, respectively) hypoperfusion was less pronounced with repetitive SDs. The three data points (proximal values ~0.3 ipsi/contra from one sham animal) were statistical outliers and were not included in significance tests. (b) hyperperfusion in proximal locations was consistently less than corresponding remote regions within the same animal (n = 19 SDs for both remote and proximal) and compared to proximal regions in sham mice (n = 20, 19 SDs for sham and dMCAo, respectively). For data shown in both (a) and (b), the third SD stimulation in one sham animal took longer than 20 min. and this SD was thus excluded. In one animal from the dMCAo group, imaging artifacts emerged during the second SD stimulation which prevented analysis of perfusion changes in both remote and proximal ROIs for this SD. In a second dMCAo animal, there were no detectable perfusion changes (or DC shifts detected in this same animal, see Figure 2) during the third SD in both remote and proximal locations, despite SD propagation confirmation near the induction site. Two-way ANOVA for hypoperfusion data (a): group (sham vs. stroke), F (DFn, DFd) = F(1,71) = 3.943, P = 0.0509; ROI location (remote vs. proximal), F(DFn, DFd) = F(1,71) = 43.24, P<0.0001; interaction, F(DFn, DFd) = F(1,71) = 4.724, P = 0.0331. Two-way ANOVA for hyperperfusion data (b): group (sham vs. stroke), F (DFn, DFd) = F(1,74) = 143.3, P <0.0001; ROI location (remote vs. proximal), F(DFn, DFd) = F(1,74) = 45.62, P<0.0001; interaction, F(DFn, DFd) = F(1,71) = 54.33, P <0.0001. * P<0.05, **** P<0.0001 from Bonferroni’s multiple comparisons test.

ABBREVIATIONS

Ca²⁺

calcium

CBF

cerebral blood flow

CSF

cerebral spinal fluid

DC

direct current

dMCAO

distal middle cerebral artery occlusion

ICU

intensive care unit

i.p.

intraperitoneal

KCl

potassium chloride

K⁺

potassium

LSCI

laser speckle contrast imaging

NMDA

N-methyl-D-aspartate

SD

spreading depolarization