Transient selective brain cooling confers neurovascular and functional protection from acute to chronic stages of ischemia/reperfusion brain injury

Abstract

Ischemic injury can be alleviated by the judicious use of hypothermia. However, the optimal regimens and the temporal kinetics of post-stroke neurovascular responses to hypothermic intervention have not been systematically studied. These gaps slow the clinical translation of hypothermia as an anti-stroke therapy. Here, we characterized the effects of transient selective brain hypothermia (TSBH) from the hyperacute to chronic stages of focal ischemia/reperfusion brain injury induced by transient middle cerebral artery occlusion in mice. A simple cooling device was used to induce TSBH during cerebral ischemia. This treatment reduced mortality from 31.8% to 0% and improved neurological outcomes for at least 35 days post-injury. TSBH mitigated blood–brain barrier leakage during the hyperacute and acute injury stages (1–23 h post-reperfusion). This early protection of the blood–brain barrier was associated with anti-inflammatory phenotypic polarization of microglia/macrophages, reduced production of pro-inflammatory cytokines, and less brain infiltration of neutrophils and macrophages during the subacute injury stage (three days post-reperfusion). TSBH elicited enduring protective effects on both grey and white matter for at least 35 days post-injury and preserved the long-term electrophysiological function of fiber tracts. In conclusion, TSBH ameliorates ischemia/reperfusion injury in the neurovascular unit from hyperacute to chronic injury stages after experimental stroke.

Introduction

Therapeutic hypothermia has been in clinical use for more than five decades and is one of the most robust neuroprotective interventions against ischemic stroke discovered to date. Hypothermia is often described as “suspended animation,” as it slows or stops metabolism without killing the organism.¹ Mechanistic studies indicate pleiotropic mechanisms underlying hypothermia-afforded protection of the brain after ischemic injury, including preservation of ATP, maintenance of pH, and reduction of lactate and glutamate buildup.¹ Furthermore, nearly every cell death pathway examined to date appears to be inhibited by hypothermia. An ongoing multicenter, randomized, Phase III clinical trial is testing the therapeutic efficacy of hypothermia in ischemic stroke patients, based on promising results of a Phase II clinical trial (ClinicalTrials.gov Identifier: NCT01833312).² Furthermore, a series of positive results from clinical trials of recanalization therapy provide an important opportunity for exploiting the restorative effects of early reperfusion.^3,4 Thus, the potential value of a combination therapy of hypothermia and recanalization during acute stroke stages is worthy of further scientific inquiry.^5,6

Concerns about the feasibility and efficacy of therapeutic hypothermia remain the biggest obstacle to its clinical translation. Furthermore, hypothermia regimens have not been standardized, and the target temperature and therapeutic duration of hypothermia still need to be optimized.^7–9 Recent studies suggest that prolonged cooling time of 24 to 48 h might be essential for long-term benefits of therapeutic hypothermia.¹⁰ However, long periods of hypothermia increase the incidence of serious complications and patient discomfort.^11,12 Furthermore, targeted hypothermia of the brain may lead to fewer side-effects than whole-body cooling. Thus, to accelerate the clinical translation of therapeutic hypothermia, it is important to avoid unnecessary prolongation of hypothermia duration and to test whether transient hypothermia of the brain alone elicits acute and long-lasting benefits in experimental stroke.

The major goals of the present study were to test whether transient selective brain hypothermia (TSBH) would protect the neurovascular unit in both acute injury stages and long-term recovery stages after transient middle cerebral artery occlusion (tMCAO) in mice. A physical cooling device was placed on the mouse brain to avoid the complications of systemic hypothermia. The spatiotemporal characteristics of neuronal tissue loss, blood–brain barrier (BBB) disruption, immune cell infiltration, chemokine/cytokine secretion, white matter injury, and neurological outcomes were analyzed at various time points, ranging from the hyperacute (1 h) to the acute injury phase (hours), the subacute injury phase (days), and the chronic repair phase (weeks). Previous hypothermia studies typically examined short time periods only and did not confirm long-lasting therapeutic effects. Therefore, it is still unclear whether hypothermia is only transiently protective and merely delays injury.⁸ Here we report that TSBH preserves the integrity of the BBB in hyperacute to acute injury stages, and that this is associated with the expected reduction in immune cell infiltration, lower chemokine/cytokine levels, and anti-inflammatory phenotypic polarization of microglia/macrophages in subacute phases after tMCAO in mice. This early shift in immune status shortly after reperfusion is linked with long-lasting protection of grey and white matter, improved electrophysiological function of white matter tracts, and superior neurological recovery in the chronic/repair phase after ischemia/reperfusion brain injury.

Materials and methods

Animal model

All animal experiments were performed at the University of Pittsburgh and were approved by the University of Pittsburgh Institutional Animal Care and Use Committee. We strictly followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and the experiments reported are in compliance with the ARRIVE guidelines. Male 8–10-weeks-old C57BL/6J mice were purchased from The Jackson Laboratory. All mice were housed in a specific pathogen-free facility with a 12-h light/dark cycle at the University of Pittsburgh. Mice were randomly assigned to normothermia, hypothermia, or sham control groups with a lottery drawing box. Mice were anesthetized with 1.5% isoflurane in a 30% O₂/68.5% N₂O mixture. tMCAO was induced in the left hemisphere for 60 min as described in pervious publications.¹³ Laser Doppler flowmetry was used to confirm regional cerebral blood flow (CBF) reduction to >70% of baseline values during MCAO. Surgeries and all stroke outcome assessments were performed by investigators blinded to experimental group assignments.

Hypothermia and temperature monitoring

Local hypothermia was induced under anesthesia starting at 10 min after the onset of ischemia. The cooling pad was constructed of tinfoil (12-mm width) and cooled on dry ice in an insulated storage container. The size of the cooling pad and the cooling time were kept consistent and the temperature of the cooling pad was maintained at approximately 0℃. The cooling pad was placed on the skull over the ischemic hemisphere on a sterile drape. Body temperature was measured continuously through the rectum and maintained at 36.5 to 37.5℃ with a regulated heating pad (Physitemp; TCAT-2LV controller). In a pilot study, the temperatures of the striatum of the forebrain and the sub-temporalis muscle were measured with a thermometer (Supplementary Figure 1). One temperature probe was inserted to a depth of 3.0 mm at 0.86 mm anterior to Bregma and 2 mm lateral to the sagittal suture, to assess striatal temperature. At the same time, another probe was inserted into the gap between the temporalis and the skull. The baseline temperature and the temperatures during hypothermia and rewarming were all recorded. In order to achieve the target temperature in the striatum (between 35 and 36℃), we discovered that the temperature of the sub-temporalis muscle had to be maintained between 32 and 34℃ (Supplementary Figure 1). Thus, in subsequent experiments, only the sub-temporalis temperature was monitored, without insertion of an invasive probe directly into the brain.

Two-dimensional laser speckle imaging

Regional CBF was monitored before ischemia, after the onset of ischemia (10 min after MCAO), after hypothermia/normothermia (60 min after MCAO), and after reperfusion (10 min and 60 min after reperfusion), using the laser speckle imaging technique, as previously described.¹⁴ Changes in CBF were expressed as a percentage of pre-MCAO baseline values. The ischemic core area was defined as the area in which CBF decreased by more than 70% of the baseline values. The penumbra was defined as the area in which CBF decreased more than 50% but less than 70% of the baseline values.

Assessments of BBB disruption after MCAO

BBB permeability was assessed by measuring the brain extravasation of a fluorescent tracer (Alexa 555 cadaverine) injected intravenously 1 h before sacrifice, as described previously.¹³ Endogenous plasma IgG leakage into brain parenchyma was also used as a marker of BBB permeability. Mice were deeply anaesthetized and transcardially perfused with 0.9% NaCl followed by 4% paraformaldehyde in PBS. Brains were collected and cryoprotected in 30% sucrose in PBS, and frozen serial coronal brain sections (25 µm-thick) were prepared on a cryostat (CM1900, Leica, Bensheim, Germany). Brain sections were processed for direct fluorescent detection of Alexa 555 or immunofluorescent staining of IgG as described below. Briefly, sections were blocked by the Avidin/Biotin Blocking Kit (SP-2001; Vector Labs) for 15 min followed by 5% donkey serum for 60 min. Sections were then incubated with the biotinylated anti-mouse IgG reagent from the MOM™ kit (BA-2000; Vector Labs), followed by the Alexa 488-streptavidin secondary antibody (016-540-084; Jackson ImmuneResearch). Images were acquired using an inverted Nikon Diaphot-300 fluorescence microscope equipped with a SPOT RT slider camera and Meta Series Software 5.0 (Molecular Devices, Sunnyvale, CA, USA). Six equally spaced sections encompassing the MCA territory were quantified for cross-sectional area of cadaverine or IgG fluorescence by an investigator blinded to treatment groups using ImageJ software. These areas were summed and multiplied by the distance between sections (1 mm) to yield a leakage volume in mm³.

Neurological function evaluations

As described previously,¹⁵ the accelerating rotarod test (motor coordination), adhesive removal test (sensorimotor function), foot fault test (sensorimotor function), and Morris water maze test (spatial learning and memory) were performed to assess neurological functions before and 3, 5, 7, 10, 14, 21, 28, 35 days after MCAO. The mice that died before 35 days after tMCAO were excluded from the final behavioral analyses. In the first three days after MCAO, neurological functions were evaluated by a six-point scale as follows: 0: No visible neurological deficits. 1: Failure to extend right forepaw completely, suggesting mild focal neurological deficits. 2: Circling to the right, suggesting moderate focal neurological deficits. 3: Falling to the right, indicating severe focal neurological deficits. 4: No spontaneous walking and depressed level of consciousness. 5: Unresponsive to stimulation or death due to brain ischemia.

Real-time polymerase chain reaction

Total RNA was extracted from ischemic brains three days after stroke using the RNeasy Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. One microgram of RNA was used to synthesize the first strand of cDNA using the Superscript First-Strand Synthesis System for Real-Time Polymerase Chain Reaction (RT-PCR) (Invitrogen, Carlsbad, CA). PCR was performed on the Opticon 2 Real-Time PCR Detection System (Bio-Rad, Hercules, CA) using corresponding primers and SYBR Green PCR Master Mix (Invitrogen). The cycle time values were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) levels within the same sample. The expression levels of mRNAs were then expressed as fold-changes versus sham control. The primer sequences are provided in Supplementary Table 1. The melting curve of the sequences was verified in five independent experiments.

Immunohistochemistry

Immunohistochemistry was performed on 25-µm-thick coronal brain slices, as described previously.¹³ Primary antibodies included the following: rabbit anti-microtubule-associated protein 2 (MAP2; sc-20172; Santa Cruz); goat anti-Iba1 (ab5076; Abcam); rat anti-CD16 (553142; BD pharm); goat anti-CD206 (AF2535; R&D); rabbit anti-beta amyloid precursor protein (β-APP; 512700; Fisher Scientific); mouse anti-200kD neurofilament heavy (NF200; MAB5262; MilliporeSigma); rabbit anti-myelin basic protein (MBP; ab5622; Abcam); mouse anti-SMI-32 (ab50761; Abcam); rabbit anti-CCR2 (ab32114; Abcam). Nonspecific antibody binding was blocked by incubation in 5% normal donkey serum (017-00-121; Jackson ImmunoResearch). For mouse primary antibodies, the mouse-on-mouse (M.O.M) blocking reagent (MKB-2213; Vector Laboratories) was applied prior to adding mouse primary antibodies. Donkey secondary antibodies conjugated with DyLight 488 or Cy3 (Jackson ImmunoResearch) were used.

Flow cytometry

Three days after MCAO or sham surgery, the ipsilateral hemisphere was collected and single cells were extracted from the homogenate. The extraction protocol was specifically designed for lymphocytes and monocytes.¹⁵ Briefly, 30% and 70% Percoll gradients (GE Healthcare BioSciences) were produced to eliminate myelin and cell debris. After centrifugation at 500 × g for 30 min (without acceleration or braking), the complete single cells were located at the interface of 30% and 70% Percoll. The extracted single cells were gently washed with FACS buffer (1% PS antibiotic, 2 nM EDTA, 2% FBS in HBSS buffer), and pre-validated antibodies were added into each single cell sample and incubated for 30 min on ice in the dark. The antibodies used were as follows: CD19-BUV395 (BD 563557); CD3-FITC (eBioscience 11-0032-82); CD45-PerCPcy5.5 (eBioscience 45-0451-82); CD11b-BUV737 (BD 564443); F4/80-BV605 (BD 743281); Gr1-PE (eBioscience 12-5931-81); CD16-PE-Cy7(BD 560829); CCR2-BV421 (BioLegend 150605); PE-Cy7 Rat IgG2b, κ Isotype control (BD 552849). Flow cytometry was performed on the BD LSRII flow cytometer (BD Biosciences) according to the manufacturer's instructions. Data were analyzed with Flow Jo software (FlowJo, LLC).

Electrophysiology

Compound actions potentials (CAPs) in the corpus callosum were measured as described previously.¹⁶ Briefly, fresh brains were harvested from mice 35 days after MCAO or sham surgery. Coronal brain slices (350 µm thick) were incubated in pre-gassed (95% O₂/5% CO₂) artificial cerebrospinal fluid (aCSF) for 0.5 h at 34℃, and left at 20℃ for 1 h. A bipolar tungsten stimulating electrode was lowered into the corpus callosum at a depth of ∼100 µm, approximately 1 mm lateral to the midline. A glass extracellular recording pipette (3 to 5 MΩ tip resistance when filled with aCSF) was lowered into the corpus callosum (0.75 mm from the stimulating electrode). Thirteen increasing intensities of the stimulus (0 µA to 2000 µA) were applied on the stimulating electrode. CAPs were acquired using pClamp 10 software (Molecular Devices), sampled at 2–10 kHz, and filtered at 1 kHz.

Statistical analyses

All analyses were performed by investigators blinded to treatment groups. Data are presented as mean ± standard deviation (SD), following tests for normality and homogeneity of variance. Individual data points are plotted where applicable. For data from two groups with Gaussian distributions and equal variances, the Student's t-test (two-tailed) was performed. For non-Gaussian data or data with unequal variances, nonparametric tests were used. Data for neurobehavioral tests were analyzed by a two-way repeated measures (RM) analysis of variance (ANOVA) followed by the Bonferroni post hoc correction. The Pearson product linear regression analysis was used to correlate the multiple histological parameters and sensorimotor behaviors. A p value less than 0.05 was deemed statistically significant. Supplementary Table 2 lists additional details about the statistical analyses.

Results

Transient and selective brain hypothermia does not increase regional CBF during or immediately after cerebral ischemia in mice

To induce evenly distributed and selective hypothermia of the brain, we measured the temperature of the rectum (core body temperature), the striatum (brain temperature), and the sub-temporalis muscle (local temperature) following application of the cooling device in a pilot experiment. A robust correlation between sub-temporalis muscle temperature and brain temperature was confirmed (Supplementary Figure 1). According to the temperature monitoring, the cerebral temperature was lowered within 5–10 min after application of the cooling device compared to the normothermia group, and this loss in temperature endured for the subsequent 50 min until the onset of rewarming (Figure 1(a)).

Figure 1.

Transient selective brain cooling acutely decreases cerebral temperature during ischemia without improving cerebral blood flow in a murine model of stroke. Mice were subjected to normothermia (NT) or hypothermia (HT) during transient middle cerebral artery occlusion (tMCAO). (a) Shown is the schematic diagram of the experimental design and procedures. Dynamic temperature monitoring of the striatum during tMCAO and reperfusion in pilot experiments are shown in the upper right panel. Data are shown as mean ± SD. N=3 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001 HT vs. NT. (b–d) Regional cerebral blood flow (CBF) was monitored using two-dimensional laser speckle imaging techniques. (b) Representative images and quantification of CBF before, during, and after tMCAO. (c) Representative images and quantification of surface areas where the CBF decreased by more than 70% of baseline (blue; defined as ischemic core) at the indicated time points after ischemia or reperfusion onset. (d) Representative images and quantification of surface areas where the CBF decreased by more than 50% but less than 70% of baseline (light blue; defined as penumbra) at the indicated time points. The ischemic core areas are in grey. Data are shown as mean ± SD. N=12 mice per group. **p < 0.01 HT vs. NT. ns implicates no significant difference.

As blood flow might also be affected by the cooling device and by brain temperature, CBF was monitored during ischemia and reperfusion. There was no significant difference between hypothermia and normothermia groups in pre-ischemic CBF baseline values, or in CBF changes at 10 min after ischemia (before hypothermia), 60 min after ischemia (50 min after hypothermia), 10 min after reperfusion/rewarming, or 60 min after reperfusion/rewarming (Figure 1(b)).

To further investigate the effects of TSBH on regional CBF, the sizes of the ischemic core (defined as areas with CBF decrease of at least 70% compared to baseline) and penumbra (defined as areas with 50–70% of CBF decrease) were measured (Figure 1(c) and (d)). The ischemic core area—as defined by CBF—was approximately one-fold larger in the TSBH group than the normothermia group (Figure 1(c)). However, this CBF difference disappeared by 10 min after reperfusion and thereafter. The sizes of penumbra were comparable between TSBH and normothermia groups in all time points monitored (Figure 1(d)). In summary, any protective effects of the head cooling procedure could not be attributed to simple improvements in regional CBF during either tMCAO or reperfusion.

Transient and selective brain hypothermia facilitates sensorimotor and cognitive recovery in both early and late stages after experimental stroke

Approximately 31.8% of normothermia control mice died within the first 10 days after tMCAO (Figure 2(a)). Remarkably, no mice in the hypothermia group died at any time point examined. To determine whether transient hypothermia could promote the long-term recovery of neurological functions, a battery of behavior tests was performed up to 35 days after MCAO. Brain ischemia and reperfusion induced prominent sensorimotor deficits in the normothermia control mice, which persisted for at least 35 days after MCAO. Importantly, mice with TSBH exhibited significant improvements in sensorimotor functions as well as accelerated rates of recovery compared to normothermia mice, according to the foot fault test (Figure 2(b); F_(2,22)=26.105, p < 0.001 by two-way RM ANOVA; post hoc p < 0.001 HT MCAO vs. NT MCAO), rotarod test (Figure 2(c); F_(2,29)=16.516, p < 0.001 by two-way RM ANOVA; post hoc p = 0.001 HT MCAO vs. NT MCAO), and adhesive removal test (Figure 2(d); contact time F_(2,29)=10.104, p < 0.001 by two-way RM ANOVA; post hoc p = 0.001 HT MCAO vs. NT MCAO; removal time F_(2,29)=8.698, p = 0.001; post hoc p = 0.002). The improvements were apparent within three to seven days post-injury and appeared long-lasting. Aside from sensorimotor functional improvements, long-term benefits of hypothermia on cognition were also observed in the spatial learning phase of the Morris water maze test (Figure 2(e) and (f); F_{(2, 29)}=7.656, p = 0.0049 HT MCAO vs. NT MCAO). Furthermore, mice in the hypothermia group spent more time in the goal quadrant trying to find the platform in the probe test, indicating superior spatial memory compared to normothermia mice (Figure 2(e) and (g)). Consistent with these behavioral indicators, the hypothermia group also exhibited dramatic protection against brain tissue loss at 35 days after tMCAO, as assessed by immunostaining for the neuronal marker MAP2 (Figure 2(h)). These collective results confirm that TSBH elicits enduring positive effects upon organismal survival, neurological functions, rate of spontaneous behavioral recovery, and brain tissue preservation after ischemia/reperfusion brain injury in vivo.

Figure 2.

Targeted selective brain hypothermia applied during ischemia reduces mortality, neurological dysfunction, and tissue loss in the subacute and/or chronic stages of experimental stroke injury. Mice were subjected to sham operation or 1 h of MCAO. Normothermia (NT) or hypothermia (HT) was induced during MCAO, and stroke outcomes were assessed. (a) Survival curve up to 35 days after MCAO. N=22 mice for NT. N=14 mice for HT. (b–d) Sensorimotor behavior tests were performed on sham-operated mice (under normothermia) or mice subjected to NT or HT during tMCAO, before and 3–35 days after surgery. (b) Foot fault test. N=4 mice for sham. N=7 mice for NT. N=14 mice for HT. (c) Rotarod test. N=4 mice for sham. N=14 mice for NT and HT groups. (d) Adhesive removal test. N=4 mice for sham. N=14 mice for NT and HT groups. (e–g) Morris water maze test. Cued tests were performed 22–26 days after MCAO or sham operation to evaluate spatial learning. Probe tests were performed on the day following the end of cued tests (27 days after tMCAO or sham operation) to assess spatial memory. Representative trail graphs (e) and quantification (f and g) of cued and probe tests are shown. N=4 mice for sham. N=14 mice for NT and HT groups. (h) Loss of grey matter was evaluated in coronal brain sections immunostained for the neuronal marker MAP2 (green) at 35 days after tMCAO. N=12 mice per group. Data are shown as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001 HT vs. NT.

Transient and selective brain hypothermia prevents early BBB hyperpermeability shortly after reperfusion in experimental stroke

The improvement of sensorimotor functions in TSBH-treated mice within three days after tMCAO (Figure 2(b) and (d)) suggested that TSBH may provide early protection against ischemia/reperfusion brain injury. Therefore, we examined BBB permeability and stroke outcomes in the first three days after tMCAO. The impairment of the BBB after tMCAO was identified at 1–23 h post-reperfusion by the leakage of an intravenously injected fluorescent molecule Alexa 555 cadaverine (3 kDa) as well as endogenous plasma IgGs (∼150 kDa) into brain parenchyma (Figure 3(a)). BBB integrity was robustly preserved by hypothermia therapy from 1 h to at least 23 h after reperfusion (Figure 3(a) and (b)). Neurological deficit scores and body weight losses were also significantly lower in the hypothermia group in the first three days after tMCAO (Figure 3(c); neurological deficit score F_(1,10)=32.704, p < 0.001 HT MCAO vs. NT MCAO by two-way RM ANOVA; body weight loss F_(1,10)=5.758, p = 0.037 HT MCAO vs. NT MCAO by two-way RM ANOVA). Hypothermia dramatically reduced the infarct volume at three days after ischemia (Figure 3(d)), consistent with the long-term effects reported in Figure 2. These results confirm that hypothermia preserves the integrity of the BBB shortly after reperfusion onset during acute injury stage, and that this is associated with an early and dramatic reduction in grey matter tissue loss and neurological deficit scores.

Figure 3.

Transient selective brain hypothermia preserves the integrity of the blood–brain barrier during the hyperacute and acute stages of experimental stroke injury. Mice were subjected to normothermia (NT) or hypothermia (HT) during MCAO. (a) Representative images to illustrate the brain extravasation of the small molecule 555-cadaverine (3 kDa; red) and endogenous IgG molecules (∼150 kDa; green) at three time points (1 h, 3 h and 23 h) after reperfusion onset in NT and HT groups subjected to tMCAO. (b) The volume with cadaverine or IgG leakage was calculated at 1 h, 3 h, and 24 h of reperfusion. (c) The neurological deficit score and loss of body weight (compared to pre-surgery body weight) in the first three days after tMCAO in the NT and HT groups. (d) Loss of grey matter was evaluated in coronal brain sections immunostained for the neuronal marker MAP2 (green) three days after tMCAO. Data are shown as mean ± SD. N=6 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001 HT vs. NT.

Transient and selective brain hypothermia prevents immune cell infiltration and suppresses pro-inflammatory reactions in the brain after experimental stroke

Early BBB damage may contribute to permanent neurovascular impairments by facilitating the infiltration of immune cells into the injured nervous system and their pro-inflammatory actions.¹³ Therefore, we tested the hypothesis that TSBH would mitigate the post-injury inflammatory milieu. The infiltration of peripheral immune cells into the brain was assessed with flow cytometry (Supplementary Figure 2(a)) at three days after tMCAO. Ischemia/reperfusion prominently elevated the number of CD45^highF4/80^-Gr1⁺ cells (neutrophils) and CD45^highF4/80⁺ cells (macrophages and activated microglia) in the ipsilateral brain hemisphere three days after tMCAO in the normothermia group (Figure 4(a); neutrophil F_(3,15)=6.038, p = 0.007 by one-way ANOVA, post hoc p = 0.025 NT MCAO vs. NT sham; macrophage F_(3,15)=28.897, p < 0.001 by one-way ANOVA, post hoc p < 0.001 NT MCAO vs. NT sham). Importantly, TSBH dramatically reduced the numbers of these cells in the brain (Figure 4(a); neutrophil p = 0.024 HT MCAO vs. NT MCAO; macrophage p < 0.001 HT MCAO vs. NT MCAO), suggesting a reduction of neutrophil and macrophage infiltration into the post-stroke brain. Aside from a reduction in neutrophil and macrophage invasion of the brain in the hypothermia group, a higher percentage of microglia in the brain exhibited the resting phenotype (CD11b⁺CD45^int) in the hypothermia group (Figure 4(a)). Hypothermia did not alter the brain infiltration of lymphocytes after MCAO, including CD11b⁻CD45^highCD3⁺CD19⁻ T cells and CD11b⁻CD45^highCD3⁻CD19⁺ B cells (Supplementary Figure 2(b)). Since the migration of circulating immune cells toward injury sites relies on their recognition of various chemokines released by injured tissue, we examined mRNA expression of a variety of chemokines in the brain at three days after tMCAO. Stroke-induced expression of a number of critical chemokines, including CCL2, the ligand of CCR2, was decreased in the hypothermia group compared to normothermic control mice (Figure 4(b) and Supplementary Figure 3). Out of all 10 chemokines measured, only CCL21 levels were not significantly decreased by hypothermia.

Figure 4.

Hypothermia blunts the recruitment of specific immune cell populations and tempers pro-inflammatory cytokine release during the subacute stage of experimental stroke injury. Mice were subjected to normothermia (NT) or hypothermia (HT) during MCAO or sham operation. (a) The numbers of infiltrated neutrophils and macrophages, and resident microglia were quantified in the ipsilateral hemisphere by flow cytometry at three days after MCAO or sham operation. Shown are representative density plots (left panel) and summarized graphs (right panel). No difference was observed between NT sham and HT sham groups; therefore, only the density plot of NT sham is shown. N=4 mice for NT sham. N=3 mice for HT sham. N=6 mice for NT and HT MCAO groups. (b) mRNA levels of a panel of chemokines were measured in the ipsilateral hemisphere by RT-PCR at three days after MCAO or sham operation. N=6 mice per group. (c) mRNA levels of pro-inflammatory cytokines interleukin (IL)-1α, tumor necrosis factor (TNF)-α, IL-6, and inducible nitric oxide synthase (iNOS) were measured in the ipsilateral hemisphere by RT-PCR at three days after MCAO or sham operation. N=6 mice per group. (d) Expression of CCR2 was examined in flow cytometry-gated cells at three days after MCAO in NT mice. Left panel: Representative CCR2 expression panel in the CD11b⁺CD45^high cell population (red; including neutrophils, macrophages, and activated microglia) and CD11b⁺CD45^int cell population (blue; microglia) in the brain. Right panel: Comparison of CCR2 expression in the CD45^highF4/80⁺ population (macrophages) harvested from spleen, blood, or brain. N=3 mice. The dash line indicates the cutoff line between the CCR2⁻ and CCR2⁺ populations. Data are shown as mean ± SD. ^#p < 0.05, ^##p < 0.01, ^###p < 0.001 NT MCAO vs. NT sham. *p < 0.05, **p < 0.01, ***p < 0.001 HT MCAO vs. NT MCAO.

Immune cell recruitment into the brain is expected to increase the secretion of pro-inflammatory cytokines. Therefore, a panel of cytokines (IL-1α, TNF-α, IL-6, and iNOS) were measured in the ipsilateral brain hemisphere by RT-PCR at three days after tMCAO. As expected based on preservation of the BBB and reduced immune cell infiltration in the hypothermia group, TSBH abolished the increase in each of these pro-inflammatory cytokines after ischemic injury (Figure 4(c)).

To further verify the relationship between immune cell infiltration and chemokine production, we examined the CCL2-CCR2 coupling. In the post-MCAO brain, compared with CD11b⁺CD45^int microglia, the majority of cells in the CD11b⁺CD45^high population (neutrophils, macrophages, and activated microglia) expressed CCR2 on the cell surface (Figure 4(d)). Furthermore, characterization of macrophages (CD45^highF4/80⁺ cells) in different organs revealed that only a small portion of splenic macrophages (17.46%) and blood macrophages (35.63%) expressed CCR2—far lower than the CCR2⁺ macrophage cell densities in the ischemic brain (86.37%). These patterns of CCR2 expression in macrophages of the spleen, blood, and injured brain support a role for CCR2-CCL2 coupling in macrophage infiltration into the brain. CCR2 expression on brain macrophages/activated microglia was further confirmed by double-label Iba1/CCR2 immunostaining at three days after MCAO (Supplementary Figure 4(a)).

Transient and selective brain hypothermia encourages polarization of microglia and macrophages towards an anti-inflammatory phenotype after ischemia/reperfusion injury

As hypothermia therapy reduced macrophage infiltration into the brain after ischemia, we explored the relationship between brain microglia/macrophages and the protective effects of hypothermia. Iba1/CD16 and Iba1/CD206 double-label immunostaining was performed to label different subpopulations of brain microglia/macrophages at three days after ischemic stroke. In the normothermia group, ischemia/reperfusion robustly elevated the numbers of Iba1⁺CD16⁺ cells and Iba1⁺CD206⁺ cells in the peri-infarct area compared to sham-operated mice (Figure 5(a), Supplementary Figure 4(b) and (c)). Compared to the normothermia group, mice in the TSBH group exhibited fewer Iba1⁺CD16⁺ cells three days after tMCAO, which are likely the pro-inflammatory microglia/macrophages (Figure 5(a) and (b)).¹⁷ On the contrary, higher numbers of Iba1⁺CD206⁺ cells were observed in the TSBH group than normothermia group, which are likely the anti-inflammatory microglia/macrophages (Figure 5(a) and (b)). The numbers of total Iba1⁺ cells were comparable between hypothermia and normothermia groups (Figure 5(b)). Flow cytometry data confirmed that three days after tMCAO, hypothermia group had fewer CD16⁺ cells in the CD45^highF4/80⁺ cell population (macrophages and activated microglia) in the brain, compared to the normothermia group (Figure 5(c)). The decrease of brain Iba1⁺CD16⁺ cells in hypothermia-treated mice lasted for at least seven days post-MCAO (Figure 5(d)).

Figure 5.

Hypothermia promotes a shift towards anti-inflammatory phenotypes in microglia/macrophages in the subacute stage after experimental stroke injury. Mice were subjected to normothermia (NT) or hypothermia (HT) during MCAO. (a and b) Three days after MCAO in NT or HT conditions, coronal brain sections were stained for the microglial/macrophage marker Iba1 and the pro-inflammatory like phenotypic marker CD16 or the anti-inflammatory like marker CD206. (a) Representative images of double-label Iba1/CD16 and Iba1/CD206 immunofluorescence in the ipsilateral peri-infarct regions (see upper panel for illustration) are shown. (b) Left panel: The quantification of total Iba1⁺ cells. Right panel: The numbers of Iba1⁺/CD16⁺ cells and Iba1⁺/CD206⁺ cells were quantified and expressed as numbers over 100 Iba1⁺ cells. N=4 mice per group. (c) Flow cytometry of CD16 expression in the CD45^highF4/80⁺ macrophage/activated microglia population. Left panel: representative histograms showing the number of CD16⁺ cells in HT and NT groups. The dash line indicates the cutoff line between the CD16⁻ and CD16⁺ populations. Upper right panel: CD45^highF4/80⁺ cells were displayed on a CD16/FSC density plot. Images of NT and HT groups are merged, showing the HT brain (red dots) had less CD16⁺ cells (square) than the NT brain (grey dots). Quantification of CD16⁺ cell number per 1000 cells was shown in the lower right panel. N=5 mice for NT. N=7 mice for HT. (d) Seven days after MCAO, brain sections were immunostained for Iba1 and CD16 and the number of double-positive cells was quantified. N=6 mice per group. (e) Seven days after MCAO or sham operation, brain sections were double-stained for beta amyloid precursor protein (β-APP; marker of axonal damage) and NF200 (axon marker) in the ipsilateral striatum (STR) and external capsule (EC). Arrows: the axonal bundles in the STR that are enlarged in the insets. Yellow arrowheads: β-APP accumulation in the STR and EC. The number of β-APP per mm² was quantified. N=5–6 mice per group. (f) Pearson correlation between the striatal β-APP accumulation and the number of CD16⁺/Iba1⁺ cells (upper panel) or animals' performance in the rotarod test at seven days after tMCAO (lower panel). N=6 mice per group. Data are shown as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001 HT vs. NT. ns implicates no significant difference.

As a marker of axonal damage, β-APP accumulation was measured in the striatum and external capsule at seven days post-ischemia. MCAO-induced accumulation of β-APP was almost completely abolished by TSBH (Figure 5(e)). Interestingly, these measurements of axonal damage demonstrated a strong positive correlation with the number of brain Iba1⁺CD16⁺ cells (pro-inflammatory microglia/macrophages), and a negative correlation with the animals' performance in the accelerating rotarod test at seven days after tMCAO (Figure 5(f)), suggesting that pro-inflammatory microglia/macrophages may contribute to the development of axonal injury and functional deficits after ischemia/reperfusion brain injury.

Transient and selective brain hypothermia promotes white matter integrity after ischemia/reperfusion injury

Thus far, we had shown hypothermia-induced reductions in neuronal tissue loss and preservation of axonal integrity after tMCAO. Therefore, we also determined if these beneficial effects extended to white matter. Demyelination was evaluated by calculating the ratio of SMI-32 staining of perilesional region to the levels of myelin basic protein (MBP; Figure 6(a)).¹⁶ The increase in SMI-32/MBP ratios reflects robust white matter injury after tMCAO and was abolished in the hypothermia group in both the cortex and striatum (Figure 6(b)). Furthermore, the SMI-32/MBP ratio was inversely correlated with sensorimotor performance in the rotarod test at 35 days after tMCAO (Figure 6(c)).

Figure 6.

Hypothermia improves the long-term structural integrity of white matter in the chronic stages of experimental stroke injury. Mice were subjected to normothermia (NT) or hypothermia (HT) during MCAO. White matter injury was assessed at 35 days after MCAO by double-label immunofluorescent staining of myelin basic protein (MBP) and the demyelination marker SMI-32. (a) Representative images showing MBP and SMI-32 immunofluorescence in the ipsilateral peri-infarct cortex (CTX) and striatum (STR). See upper panel for illustration. Dashed line: the approximate border of the infarct. (b) The ratio of SMI-32 to MBP immunofluorescence density was calculated as a parameter for white matter injury. Data are shown as mean ± SD. ***p < 0.001 HT vs. NT. (c) Pearson correlation of SMI-32/MBP ratio in the cortex (left panel) or striatum (right panel) and the latency to fall off the rotarod on day 35 post-injury. N=4 mice for sham (under normothermia). N=8 mice for NT and HT groups.

Aside from examining loss of MBP and an increase in SMI-32, we evaluated white matter structural integrity by calculating the width of the corpus callosum and external capsule. To this end, we divided the entire corpus callosum and external capsule into 10 equal parts by tracing 11 vertical (dorsoventral) lines extending from the midline of the brain to the edge of corpus callosum and to the lateral edge of the external capsule in the contralateral and ipsilateral hemispheres (Figure 7(a)). Next, the width of the ipsilateral corpus callosum and external capsule was measured by comparing it to the intact corpus callosum on the contralateral side and expressing as a ratio of the width of the corpus callosum at the midline. Hypothermia mice exhibited a wider external capsule in the most lateral segments (Figure 7(a)).

Figure 7.

Hypothermia improves the long-term functional integrity of white matter in the chronic stages of experimental stroke injury. (a) Brain sections were immunostained for myelin basic protein (MBP) at 35 days after sham operation (under normothermia) or tMCAO in normothermic (NT) or hypothermic (HT) conditions. Each hemisphere was divided into 10 segments along the dorsoventral axis. The width of white matter fiber from the corpus callosum (CC) to the end of external capsule (EC) was measured in each of the 10 segments. To minimize inter-individual variance, the width of the fiber tracts in each of the 10 segments relative to the midline was calculated as a ratio and compared between HT and NT MCAO groups. N=4 mice for sham. N=10 mice for NT and HT MCAO groups. *p < 0.05, ***p < 0.001 NT vs. HT. (b) Brain sections were immunostained for the axon marker NF200 at 35 days after sham (continued) operation or tMCAO in NT or HT conditions. NF200 immunofluorescence intensity was calculated in the EC. Linear correlations between NF200 immunofluorescence intensity and the latency to fall off the rotarod on 35 days after tMCAO are also shown. MFI: mean florescence intensity. N=4 mice for sham. N=10 mice for NT and HT MCAO groups. *p < 0.05, **p < 0.01. (c and d) Compound action potentials (CAPs) in white matter tracts were measured ex vivo on coronal brain slices collected 35 days after MCAO or sham surgery from NT or HT mice. (c) The schematic on the left illustrates the design of the electrophysiology experiments. The red-shaded area represents the infarct. The grey and blue probes represent the stimulation and recording probes, respectively. The distance between two probes is 0.75 mm. The right two panels illustrate representative CAP curves in response to the 1000 µA stimulus. The dashed red lines frame the target positive wave induced by the nerve fiber in response to the stimulus. (d) Shown are measurements of CAP amplitude in response to increasing stimulus strength (from 0 µA to 2000 µA) in NT sham, and NT and HT MCAO groups. N=4 mice per group. *p < 0.05 HT vs. NT. Data are shown as mean ± SD.

In Figure 5, we showed an abolition of axonal damage in the hypothermia group at seven days after ischemia. We further confirmed these effects at 35 days after ischemia by assessing the immunofluorescence intensity of NF200 for intact axons in the medial segments of the external capsule (Figure 7(b)). As expected, TSBH significantly ameliorated MCAO-induced loss of NF200 in the external capsule, and NF200 immunofluorescent intensity was significantly correlated with the latency to fall off the rotarod at 35 days after tMCAO (Figure 7(b)). Considered together with the correlation between CD16⁺ cell numbers and β-APP accumulation in Figure 5(f), these results support the view that anti-inflammatory microglial/macrophage phenotypes may mitigate axonal damage in white matter and improve functional outcomes.

Thus far, we have shown that hypothermia protects the structure of white matter tracts in the striatum, cortex, corpus callosum, and external capsule. In order to confirm that the histological protection was associated with preservation of physiological function, we measured compound action potentials (CAPs) in white matter in coronal ex vivo brain sections 35 days after MCAO (Figure 7(c)). Brain ischemia/reperfusion resulted in significant reductions of CAPs in the ipsilateral corpus callosum and external capsule in the normothermia group compared to sham animals (Figure 7(d), Supplementary Figure 4(d)). Furthermore, brain slices from the normothermia group consistently exhibited lower amplitude CAPs than those of hypothermia group (Figure 7(d)), reflecting greater functional impairments in white matter in the former.

Discussion

The present findings demonstrate that transient hypothermia induced selectively to the brain starting from 10 min after the onset of tMCAO dramatically improves an array of histopathological and behavioral outcomes at acute and subacute injury stages as well as chronic recovery stages. Collectively, the data presented here reveal that the early preservation of BBB integrity shortly after reperfusion onset is associated with reduced brain infiltration of neutrophils and macrophages, dampening of pro-inflammatory chemokine and cytokine expression, polarization of microglia/macrophages towards an anti-inflammatory phenotype, reductions in structural and functional axonal damage, and grey and white matter protection in the hypothermia group. These early changes occur in the absence of any increase in regional CBF during reperfusion, and culminate in dramatically lower mortality, greater brain tissue preservation, and superior neurological outcomes at acute, subacute, and extended survival time points. Therefore, timely selective hypothermia induced by a physical cooling device during tMCAO could elicit a cascade of protective effects in the brain.

According to imaging studies, post-stroke hyperperfusion is associated with disruption of the BBB.¹⁸ It is well established that hypothermia reduces CBF in the brain with or without ischemia.¹⁹ In the present study, CBF was temporarily lowered in the ischemic core of the hypothermia group, consistent with other reports.²⁰ Previously, we showed that hypothermia mitigates post-stroke hyperperfusion.¹⁹ These observations suggest that hypothermia may exert therapeutic effects by decelerating post-stroke hyperperfusion. In addition to preventing the early disruption of the BBB after tMCAO and reducing the subsequent infiltration of neutrophils and macrophages, hypothermia downregulated the expression of chemokines that induce chemotactic mobilization of peripheral immune cells.²¹ Infiltrating peripheral neutrophils and macrophages induce secondary and irreversible BBB damage due to secretion of matrix metalloproteinases (MMPs) such as MMP-9.¹³ Hypothermia decreases MMP-9 secretion,²² and this response may also contribute to BBB protection and less immune cell infiltration. Therefore, TSBH treatment may interrupt a self-amplifying ischemic cascade that is set in motion in the acute to subacute stages of ischemic injury, encompassing early BBB disruption and subsequent immune cell infiltration and propagation of inflammation.^23,24

In most stroke studies, BBB damage is typically measured 24 h after ischemia.²⁴ However, Shi et al.¹³ demonstrated early BBB disruption during the acute stage (<3 h) after stroke, followed by secondary irreversible BBB disruption. As argued above, the acute stage of BBB disruption affects long-term outcomes by facilitating immune cell infiltration and subsequent neuroinflammatory processes.^23–25 Therefore, we evaluated BBB leakage from 1 to 23 h after reperfusion in our experiments. As expected, TSBH conferred dramatic protection against BBB breakdown from 1 to 23 h post-reperfusion onset. Subsequently, a significant decrease in the infiltration of neutrophils and macrophages into the injured brain was observed, consistent with other hypothermia reports.^26,27 Although we did not observe that TSBH suppressed the recruitment of lymphocytes, the pro-inflammatory sequelae of ischemia/reperfusion injury were dramatically alleviated in the hypothermia group. T cells and B cells contain subpopulations that may exert different or even opposing effects on pathological processes.^28–30 Further studies are warranted to test if hypothermia exerts immunosuppressive effects without decreasing the total number of lymphocytes by modifying the phenotype and function of T cell and B cell subpopulations. Alternatively, the hypothermia-induced inhibition of macrophages and neutrophils may be sufficient to reduce the expression of pro-inflammatory mediators.

A number of studies support the view that hypothermia reduces the infiltration of macrophages and drives microglia/macrophages toward an anti-inflammatory phenotype.^27,31,32 Hu et al.¹⁷ demonstrated that anti-inflammatory shifts in microglial/macrophage phenotype promote the survival of neurons with or without ischemic injury. Hypothermia therapy has also been shown to decrease the number of CD16⁺ pro-inflammatory microglia/macrophages, which are associated with white matter injury.^14,33,34 These effects are consistent with observations that pharmacological induction of whole-body hypothermia improves sensorimotor performance and diffusion tensor imaging of white matter one month after stroke.³⁵ Similarly, the protection of white matter after TSBH in our study was associated with long-lasting sensorimotor and cognitive benefits in the chronic recovery stage.

The present findings reveal robust reductions in multiple markers of injury and inflammation after application of hypothermia in stroke mice. In a number of the histological assays, we observed complete abolition of injury. Aside from the findings of the present study, hypothermia has also been shown to mitigate reactive oxygen species, glutamate-induced excitotoxicity, and apoptotic cascades such as cytochrome c release.¹ Previous studies have also shown that hypothermia protects against trauma-induced cell death in oligodendrocytes, loss of myelin, and functional impairments in neuronal circuitry.^36,37 Thus, the dramatic effects of hypothermia in the present study are probably not explained by a single factor but by its multifaceted actions.

Besides local cooling, whole-body cooling is another approach to induce therapeutic hypothermia, which can be achieved through physical methods or pharmacological activities of opioid receptor agonists, dopamine receptor activators, transient receptor potential vanilloid channel 1 agonists, etc.³⁸ Compared to physical cooling methods, pharmacological cooling may protect against brain injury through additional mechanisms other than hypothermia per se.³⁸ Like two sides of the same coin, this multi-faceted action of pharmacological hypothermia may offer more robust protection as well as unexpected toxicity and side effects.^39–41 Thus, there has been a recent trend in investigating the combination of physical and pharmacological cooling with the hope of reducing the dosage of drug while achieving comparable or even more protective effects.⁴² Nevertheless, complications may still occur during the combined treatment as a result of the systematic effect of general hypothermia, such as cardiovascular deficiency, immunodepression and infection, coagulopathy and discomfort (e.g. shivering). To this end, selective local hypothermia avoids the alteration of core temperature and lowers such potential complications. Furthermore, selective brain cooling has strong clinical implications, as a recent study demonstrated good feasibility and safety of transient intraarterial selective cooling infusion into the ischemic territory on stroke patients treated with mechanical thrombectomy.⁴³ Considering the side effects triggered by prolonged cooling time and low target temperature,⁴⁴ it is imperative to develop optimal treatment regimen with proper length and depth of therapeutic hypothermia to facilitate its clinical translation to stroke patients.

One of the limitations of the present study is that we only employed one hypothermia protocol. It is well known that the time of initiation of hypothermia can dramatically affect its therapeutic efficacy.⁴⁵ An earlier time of onset may be more important than prolonged hypothermia duration, and the robust changes in multiple markers of injury and inflammation observed in the current study probably reflect the application of hypothermia during the ischemic interval itself. Furthermore, inappropriate rewarming speed could abolish or even reverse the protection of hypothermia and elicit complications of the hematologic system.^46,47 Further studies that systematically vary the time of hypothermia onset, the target temperature, and the rewarming speed will be required for effective clinical translation.^48,49

Conclusions

By applying therapeutic hypothermia selectively to the brain, we preserved the integrity of the BBB at acute stages of stroke. This early BBB preservation was associated with subsequent inhibition of chemokine expression, fewer infiltrating neutrophils and macrophages, anti-inflammatory phenotype shifts in microglia/macrophages, lower infarct volumes, and white matter protection at subacute and/or chronic injury stages. These mechanisms collectively lead to prominent and persistent functional improvements after stroke.

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 National Key R&D Program grants #2017YFC1308401 and #2016YFC1301502 from China Ministry of Science & Technology (to X.Ji), grants #ZYLX201706 and SML20150802 from Beijing Municipal Administration (to X.Ji), and the Chang Jiang Scholar grant #T2014251 from China Ministry of Higher Education (to X.Ji). H.M. was supported by a scholarship from China Scholarship Council (201606100182). D.W. was supported in part by the National Natural Science Foundation of China (81871022). Y.S. was supported in part by the Competitive Medical Research Fund from the University of Pittsburgh Medical Center (UPMC) Health System, and a start-up fund from the Pittsburgh Institute of Brain Disorders & Recovery and Department of Neurology at the University of Pittsburgh.

This work was supported by grants from the China Ministry of Science & Technology and Beijing Municipal Administration.

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

JZ, HM and X.Ji performed experiments; JZ, LL and DW analyzed data; JZ, RKL and YS wrote the manuscript; X.Ji conceived the study and supervised the project.

Supplementary material

Supplementary material for this paper can be found at the journal website: http://journals.sagepub.com/home/jcb