Glial Cells Drive Preconditioning-Induced Blood-Brain Barrier Protection
Background and Purpose—The cerebrovascular contribution to ischemic preconditioning (IPC) has been scarcely explored. Using in vivo and in vitro approaches, we investigated the involvement of the blood-brain barrier and the role of its cellular components.
Methods—Seven-minute occlusion of the right middle cerebral artery, used as in vivo IPC stimulus 4 days before permanent occlusion of the right middle cerebral artery, significantly reduced brain infarct size (8.45±0.7 versus 13.61±0.08 mm3 measured 7 days after injury) and preserved blood-brain barrier function (Evans blue leakage, 0.54±0.1 versus 0.89±0.1 ng/mg). Assessment of neuronal, endothelial, and glial gene expression revealed that IPC specifically increased glial fibrillary acidic protein mRNA, thus showing selective astrocyte activation in IPC-protected mice.
Results—The blood-brain barrier was modeled by coculturing murine primary brain microvessel endothelial and astroglial cells. One-hour oxygen-glucose deprivation (OGD), delivered 24 hours before a 5-hour OGD, acted as an IPC stimulus, significantly attenuating the reduction in transendothelial electric resistance (199.17±11.7 versus 97.72±3.4 Ωcm2) and the increase in permeability coefficients for sodium fluorescein (0.98±0.11×10−3 versus 1.8±0.36×10−3 cm/min) and albumin (0.12±0.01×10−3 versus 0.29±0.07×10−3 cm/min) induced by severe OGD. IPC also prevented the 5-hour OGD–induced disorganization of the tight junction proteins ZO-1 and claudin-5. IPC on glial (but not endothelial) cells alone preserved transendothelial electric resistance, permeability coefficients, and ZO-1 localization after 5 hours of OGD. Astrocyte metabolic inhibition by fluorocitrate abolished IPC protection, confirming the critical role of astrocytes. IPC significantly increased glial fibrillary acidic protein, interleukin-6, vascular endothelial growth factor-a, and ciliary neurotrophic factor gene expression after OGD in glial cells, indicating that multiple pathways mediate the glial contribution to IPC.
Conclusions—Our data show that the blood-brain barrier can be directly preconditioned and that astrocytes are major mediators of IPC protection.
The concept of neuroprotection in ischemic preconditioning (IPC) is based on the observation that a brief, noninjurious stimulus protects the brain from a subsequent severe insult.1,2 Experimental IPC also has a human counterpart, as transient ischemic attacks may be associated with a better prognosis after a subsequent stroke.3,4 In vitro and in vivo IPC models have mostly focused on neurons as cellular targets of cerebral IPC, but little attention has been paid so far to the cerebrovascular compartment, even though it plays a crucial role in the pathogenesis of ischemic brain injury.5 Endothelial cells of the cerebral microvasculature are the main component of the blood-brain barrier (BBB), and their close association with astrocyte endfeet, pericytes, and microglia is essential for the maintenance of the nervous system microenvironment.6 It is known that IPC stimuli may attenuate BBB disruption and brain edema.7,8 In addition, brain endothelial cells can be preconditioned, an effect associated with stabilization of tight junction (TJ) proteins9 and/or activation of the phosphatidylinositol-3 kinase/Akt pathway.10,11 However, little is known about the contribution of glial cells to BBB preconditioning. By in vivo and in vitro approaches, this study aimed at evaluating the functional effects of IPC on the BBB and the role of its cellular components in mediating IPC-induced protection.
C57BL/6 mice (Harlan Laboratories, Italy) were housed in an Specific Pathogen Free (SPF) vivarium. Procedures involving animals and their care were conducted in conformity with institutional guidelines that are in compliance with national and international laws and policies.12
In Vivo Ischemia
Anesthesia was induced by 3% isoflurane inhalation in an N2O/O2 (70%:30%) mixture and maintained by 1% to 1.5% isoflurane inhalation in an N2O/O2 (70%:30%) mixture. Transient ischemia was induced in mice by means of a siliconized filament (7-0, Doccol Corp, Redlands, CA) introduced into the right middle cerebral artery (MCA)13,14 for 10, 7, or 5 minutes as an IPC stimulus. Severe ischemia was obtained through permanent occlusion of the right MCA (pMCAo) by electrocoagulation12 in mice anesthetized with equitensin (100 μL IP per mouse). For both surgical procedures, sham-operated mice received identical anesthesia and surgery without artery occlusion.
Quantification of Infarct Size
Twenty-micron coronal brain sections obtained from perfused brains were cut and stained with neutral red (neutral red Gurr Certistain, BDH, Poole, UK). Infarct volumes were calculated by the integration of infarct areas on each brain slice, as previously described.15
Evans Blue–Albumin Leakage
Quantitative evaluation of BBB disruption was achieved by measuring Evans blue-labeled albumin leakage in ischemic brains16 (further details are provided in supplemental Methods, http://stroke.ahajournals.org).
In Vitro BBB Model
Cocultures of brain microvessel endothelial cells, obtained from 8-week-old mice and grown on transwell inserts, and astroglial cells, obtained from 24- to 48-hour newborn mice and grown on the bottom of multiwell plates, were prepared6 (further details in supplemental Methods and supplemental Figure S1).
In Vitro Ischemia
Cocultures were exposed to oxygen-glucose deprivation (OGD; further details in supplemental Methods) according to the following protocols: 1-hour OGD (IPC group), 5-hour OGD (OGD group), or 1-hour OGD followed by 5-hour OGD 24 hours later (1-hour+5-hour OGD group) for IPC assessment. Subsequently, to discriminate between the selective contribution of endothelial and glial cells to the IPC-induced protection, the IPC stimulus (1-hour OGD) was delivered separately to either endothelial or glial cells. After the IPC stimulus, the preconditioned endothelial cells were placed together with nonpreconditioned glial cells (endo IPC), whereas nonpreconditioned endothelial cells were placed together with preconditioned glial cells (glia IPC). Twenty-four hours later, the cocultures were exposed to 5-hour OGD.
Brain microvessel endothelial cells were fixed with ice-cold ethanol for 30 minutes at 4°C. Rabbit anti–ZO-1 (1:100; Zymed, CA) and rabbit anti–claudin-5 (1:100; Zymed) were used as primary antibodies (overnight at 4°C). Anti-rabbit Alexa 594 (1:100, Molecular Probes, CA) was used as a secondary antibody. Immunofluorescence staining was assessed by using an Olympus IX81 microscope equipped with an Olympus confocal FV500 scan unit.12
Lactate Dehydrogenase Assay
Lactate dehydrogenase (LDH) activity in cell culture media was measured with a commercially available kit (CytoTox 96; Promega, Madison, WI), according to the manufacturer's instructions.
Fluorocitrate (FC; Sigma, St. Louis, MO) was prepared as described by Voloboueva et al17 to reach a final concentration of 0.48 mmol/L. Cells were exposed to FC for 3 hours, that is, 2 hours before and during 1 hour of IPC. Afterward, the medium containing the FC was replaced with fresh, normoglycemic medium, and the FC-preconditioned glial cells were cocultured together with nonpreconditioned endothelial cells.
Measurements of ATP Levels
Cellular ATP concentration was measured with the CellTiter-Glo luminescent ATP assay kit, based on the luciferase/luciferin reaction (Promega) according to the manufacturer's instructions. AVeritas luminescence counter (Turner BioSystems, Sunnyvale, CA) was used to measure the luminescence signal of the samples in white, opaque 96-well plates. ATP content was reported as luminescence units.
RNA Isolation, cDNA Synthesis, and Real-Time Polymerase Chain Reaction
Forty-eight hours after pMCAo or sham operation, the cortical mRNA expression of the following genes was analyzed18: high-molecular-weight subunit of neurofilaments and microtubule-associated protein-2 (neuronal markers); CD31 (endothelial marker); glial fibrillary acidic protein (GFAP; astrocyte marker), CD11b and CD68 (microglia/macrophage markers); and oligodendrocyte lineage transcription factor-2 and NG2 (oligodendrocyte markers). In glial cell cultures, the mRNA expression of selected genes considered to be markers of functional astrocytic phenotypes (GFAP, vascular endothelial growth factor-a, interelukin-6, and ciliary neurotrophic factor) were analyzed. The primers used are reported in supplemental Table S1 (further details in supplemental Methods).
One-way ANOVA with a Dunnett post hoc test was used for the time window of efficacy, a 1-way ANOVA with a Tukey post hoc test was used for gene expression, and an unpaired t test was used for Evans blue–albumin leakage and ATP content. Two-way ANOVA with a Bonferroni post hoc test was used for transendothelial electric resistance (TEER), and 1-way ANOVA with a Bonferroni post hoc test was used for TEER, endothelial permeability coefficients (Pe), and LDH analyses.
In Vivo IPC Model
Short, transient MCAo (10, 7, or 5 minutes) was delivered to mice as the IPC stimulus. Forty-eight hours later, mice were evaluated for neurologic deficits, ischemic volumes (by neutral red staining), and the presence of neurodegenerating cells (by fluoro-Jade staining). The 7-minute, transient MCAo (7′MCAo) was chosen as the IPC stimulus because it was the longest noninjurious stimulus1 (Supplemental Table S2).
pMCAo was delivered as severe ischemia and was performed at different time points (1, 2, 3, 4, or 5 days) after 7′MCAo (Figure 1A). When pMCAo was performed 3 or 4 days after 7′MCAo, a significant reduction in the ischemic volume, assessed 7 days later, was observed (8.95±1.02 and 8.45±0.7 mm3, respectively) compared with ischemic mice not exposed to 7′MCAo (13.61±.08 mm3; Figure 1B and 1C). Then BBB involvement in 7′MCAo-induced neuroprotection was evaluated by measuring Evans blue–albumin leakage 20 hours after pMCAo in preconditioned and nonpreconditioned mice at the time of maximal protection (4 days after 7′MCAo). Mice that had previously undergone sham occlusion (sham 7′MCAo) had marked BBB disruption 20 hours after a subsequent pMCAo (0.89±0.1 ng/mg). This was significantly attenuated in the 7′MCAo-treated group (0.54±0.1 ng/mg; Figure 1D).
To evaluate the contribution of the different cell populations of the cerebrovascular unit to the in vivo IPC, we next analyzed the cortical mRNA expression of genes that are considered to be markers of different cell types or phenotypes 2 days after pMCAo. As expected, pMCAo significantly increased CD11b, CD68, CD31, and GFAP expression compared with sham operation. pMCAo also decreased microtubule-associated protein-2 while leaving high-molecular-weight subunit of neurofilaments, NG2 and oligodendrocyte lineage transcription factor-2 unaffected (Figure 2). Previous exposure to 7′MCAo affected GFAP only, whose expression was significantly increased in preconditioned compared with nonpreconditioned mice (Figure 2).
In Vitro BBB Model
After 2 to 3 days of coculture, brain microvessel endothelial cells expressed the endothelial marker CD31 (Supplemental Figure S1). In addition, brain microvessel endothelial cells retained the features of a functional BBB6 in mice, that is, typical values of TEER (228± 6.44 Ωcm2), Pe for paracellular (sodium fluorescein Pe, 0.53±0.02×10−3 cm/min) and transcellular (albumin Pe, 0.02±0.005×10−3 cm/min) transport, and continuous intercellular staining of TJ proteins, namely, ZO-1 and claudin-5 (an intracellular and a transmembrane protein, respectively), which contribute to the tightness of the BBB–endothelial cells (supplemental Figure S1).
The experimental plan is detailed in Figure 3A. Five-hour OGD caused a drastic reduction in TEER values at all time points tested (after OGD, 93.61±3.9; 24 hours after OGD, 97.72±3.4; and 48 hours after OGD, 66.64±9.8 Ωcm2) compared with control cocultures (218±6.5, 207±6.17, and 198±6.21 Ωcm2, respectively), whereas 1-hour OGD did not affect them (after OGD, 190.59±8.2; 24 hours after OGD, 209.72±5.73; and 48 hours after OGD, 166.32±8.4 Ωcm2; Figure 3B). When 1-hour OGD was performed 24 hours before a severe 5-hour OGD, significant attenuation of the 5-hour OGD–dependent effect was observed at every time point tested (after OGD, 161.47±8.41; 24 hours after OGD, 199.17±11.7; and 48 hours after OGD, 175.84±7.11 Ωcm2; Figure 3B).
Twenty-four hours after the 5-hour OGD, a significant increase in Pe for both sodium fluorescein (1.8±0.36×10−3 cm/min) and albumin (0.29±0.07×10−3 cm/min) compared with control Pe values (sodium fluorescein, 0.53±0.02×10−3 and albumin, 0.02±0.005×10−3 cm/min) was observed (Figure 3C). Also in this case, 1-hour OGD did not significantly affect Pe values (sodium fluorescein, 0.49±0.03×10−3 and albumin, 0.09±0.002×10−3 cm/min). Cocultures exposed to 1-hour+5-hour OGD had significantly lower Pe values both for sodium fluorescein (0.98±0.11×10−3 cm/min) and albumin (0.12±0.01×10−3 cm/min) compared with those exposed to severe OGD alone (Figure 3C).
With regard to cell viability, cocultures exposed to 5-hour OGD showed a significant increase in LDH release (optical density [OD]=1.97±0.09) compared with control cells (OD= 1.1±0.08), whereas 1-hour OGD did not cause significant changes in LDH release (OD=1.44±0.08; supplemental Figure S2). Cocultures exposed to 1-hour+5-hour OGD showed a significant decrease in LDH release (OD=0.84±0.004) compared with 5-hour OGD alone (supplemental Figure S2).
The presence and distribution of TJ proteins were also assessed. Control cells showed an intense and continuous staining of ZO-1 and claudin-5 at the cell borders (Figure 3D). Five-hour OGD resulted in decreased immunostaining intensity and loss of the continuous junctional distribution, whereas 1-hour OGD did not affect them. Staining intensity and distribution of TJs in the cocultures exposed to 1-hour+5-hour OGD were similar to the control condition (Figure 3D).
Contribution of BBB Cell Populations to IPC
The experimental plan is detailed in Figure 4A. IPC delivered to glial cells alone was significantly more effective in damping the drastic reduction in TEER induced by the subsequent severe OGD, compared with IPC delivered to endothelial cells alone (Figure 4B). The effect was already visible 24 hours after severe OGD when glia IPC, but not endo IPC, showed significantly higher TEER values compared with OGD (141.44±7.62, 104.72±5.44, and 91.05±2.49 Ωcm2, respectively). Forty-eight hours after OGD, both glia IPC and endo IPC showed significantly higher TEER values compared with OGD (134.32±3.78, 110.16±2.17, and 82.88±6.05 Ωcm2, respectively; Figure 4B).
The prominent contribution of glial cells to IPC was also indicated by permeability assessment, performed 24 hours after OGD (Figure 4C). Both sodium fluorescein and albumin Pe values were significantly decreased in glia IPC (0.73± 0.07×10−3 and 0.14±0.01×10−3 cm/min, respectively) compared with OGD (2.05±1.18×10−3 and 0.34±0.04× 10−3 cm/min, respectively), whereas endo IPC showed Pe values similar to those of the OGD group (1.63±0.85×10−3 and 0.33±0.05×10−3 cm/min, respectively; Figure 4C). Regarding cell viability, both glia IPC and endo IPC showed a significant reduction in LDH release compared with OGD (OD=0.7±0.03, 0.72±0.03, and 1.88±0.02, respectively; supplemental Figure S3). The analysis of ZO-1, evaluated 48 hours after OGD, showed that its intensity and distribution in glia IPC were similar to those of controls, whereas endo IPC was similar to OGD (Figure 4D).
The role of astrocytes in mediating IPC protective effects was further assessed by exposing glial cells to the mitochondrial inhibitor FC during IPC. FC treatment significantly decreased cellular ATP content, whereas it did not affect cell viability as assessed by LDH release, showing that FC treatment was effective and not toxic (Figure 5A). The metabolic inhibition prevented the induction of IPC-induced protective effects on the BBB, as shown by the reduction in TEER values (106.4±2.5 versus 170.1±4.5 Ωcm2 at 24 hours and 95.7±5.9 versus 161.3±5.8 Ωcm2 at 48 hours; Figure 5B) and the increase in Pe values for sodium fluorescein (1.8±0.2×10−3 versus 1.3±0.1×10−3 cm/min; Figure 5C) of glia IPC FC-treated cells compared with the glia IPC group performed 24 hours after OGD. Forty-eight hours after OGD, ZO-1 intensity and distribution in glia IPC FC-treated cells were similar to those after OGD, whereas glia IPC was similar to control cells (Figure 5D).
To evaluate the molecular mechanisms involved in the prominent contribution of glial cells to IPC, immediately after OGD, we analyzed the mRNA expression of gene markers of functional astrocytic phenotypes. First, confirming our in vivo data, we observed that IPC induced a significant rise in GFAP after OGD compared with OGD alone (Figure 6). In addition, we also observed that IPC caused an increase in interleukin-6, vascular endothelial growth factor-a, and ciliary neurotrophic factor gene expression (Figure 6).
By using in vivo and in vitro approaches, this study has demonstrated the involvement of the cerebrovascular unit in IPC and emphasizes the role of astrocytes in driving the IPC-induced protective effect. The involvement of the BBB in IPC was first studied in an in vivo IPC model. According to the definition of IPC stimulus,1 we delivered 7 minutes of ischemia, representing the longest period of MCAo not associated with damage. This IPC stimulus reduced the ischemic volume (as assessed 7 days after injury) induced by a subsequent severe ischemia, and its protective effects were maximal when the IPC stimulus was delivered 4 days before pMCAo. We observed a significant reduction in BBB leakage in IPC compared with non-IPC mice. Because events occurring at the BBB as a consequence of ischemia represent major players in ischemic injury, the protective effects observed in IPC brains can be a consequence of the protective phenotype assumed by the BBB. To elucidate the selective involvement of the different brain cell populations in IPC in vivo, we analyzed gene expression for the neuronal, endothelial, and glial components. Gene expression analysis showed that only GFAP was selectively upregulated in IPC-protected mice, whereas neuronal, endothelial, microglial, and oligodendrocytic components did not appear to be involved. These findings suggest that astrocytes may have an important role in mediating BBB IPC. To further define the contribution of the BBB cell populations to IPC and to evaluate the molecular mechanisms involved in the IPC-induced protective effects, we then studied an in vitro model.
Importantly, the BBB model that we used retained the features of a functional BBB.6 Although one cannot rule out that physiologic features of the endothelial cells used in this system might be lost in vitro, findings in the present and previous studies6 indicate that these cells retain an endothelial-specific phenotype, as assessed by the expression of certain markers (that is, CD31/Platelet Endothelial Cell Adhesion Molecule-1 and claudin-5), as well as the ability to form a functionally tight barrier (that is, high TEER and low Pe values). Finally, with all the caveats of transferring in vitro data to the in vivo situation, available evidence supports the notion that astrocytes provide the cocultured endothelial cells with factors that help maintain their vascular phenotype.19
By using the in vitro BBB model, we showed that a subthreshold stimulus consisting of a short OGD protects against a subsequent severe OGD, as shown by prevention of the drastic reduction in TEER values induced by severe OGD, by the restoration of Pe values, and by the preservation of TJ organization. Notably, a 5-hour OGD did not completely disrupt TJs, otherwise TEER would have been undetectable. Rather, 5-hour OGD caused disorganization of the TJ associated with morphological changes of the cells, which became elongated and thinner compared with control cells. Under this condition, TEER and Pe values were decreased but still present.
To elucidate the contribution of endothelial and glial cells to the observed IPC-induced protective effects, the 2 BBB cell populations were exposed separately to the IPC stimulus. Preconditioned glial cells were much more efficient in eliciting the IPC protective phenotype than were endothelial cells. Specifically, when IPC was performed on glial cells alone, the BBB showed greater resistance to severe OGD, compared with the IPC performed on endothelial cells alone. It is likely that molecules produced by glial cells (either during IPC or in the following 24 hours that preceded severe OGD) are able to induce a protective BBB phenotype. Notably, the transient metabolic inhibition of astrocytes induced by FC treatment during IPC prevented the IPC protective effects on the BBB, further confirming the crucial role of astrocytes in mediating BBB protection. These results fully support the in vivo finding that astrocytes are major players in the BBB protective mechanisms activated by IPC.
Thus, this study shows for the first time that astrocytes are required for triggering an effective IPC on BBB cells. Maintenance of BBB properties is 1 of the several protective actions attributed to these cells that, with their endfeet surrounding endothelial cells on 1 side and reaching neurons on the other, are ideally situated to control metabolic supply, blood flow, ionic homeostasis, and neurotransmitter levels.20,21 Actually, glial cells have been shown to affect endothelial cells differently under different conditions. They can either worsen the ischemia-induced increase in endothelial permeability22 or protect them from pathologic stimuli, that is, from lipopolysaccharide-induced damage.23 Several mechanisms may contribute to the IPC-induced protective effect described. An increased expression of GFAP has been associated with a functional phenotype characterized by expression of specific molecules acting as inflammatory regulators, like interleukin-6, or playing a trophic role, like ciliary neurotrophic factor20 and vascular endothelial growth factor, which are endowed with protective actions on the BBB.24 Our data provide evidence that all of these molecular pathways are involved in the prominent contribution of glial cells to IPC, as shown by the significant increase in interleukin-6, vascular endothelial growth factor, and ciliary neurotrophic factor gene expression occurring after OGD.
In conclusion, we have shown that the BBB can be preconditioned and that astrocytes are a major driving gear in this mechanism, as IPC delivered to glial cells is strictly required for preventing the functional damage of the BBB. Promotion of the astrocytic signaling pathways that ultimately lead to an ischemia-tolerant phenotype may serve as a therapeutic tool for neurovascular unit protection in stroke patients.
Source of Funding
This study was partially supported by Banca Popolare di Milano.
The online-only Data Supplement is available at http://stroke.ahajournals.org/content/full/stroke.110.603266/DC1.
- Received September 21, 2010.
- Accepted January 11, 2011.
- © 2011 American Heart Association, Inc.
- Wegener S,
- Gottschalk B,
- Jovanovic V,
- Knab R,
- Fiebach JB,
- Schellinger PD,
- Kucinski T,
- Jungehulsing GJ,
- Brunecker P,
- Muller B,
- Banasik A,
- Amberger N,
- Wernecke KD,
- Siebler M,
- Rother J,
- Villringer A,
- Weih M
- Zhang Y,
- Park TS,
- Gidday JM
- Li L,
- Lundkvist A,
- Andersson D,
- Wilhelmsson U,
- Nagai N,
- Pardo AC,
- Nodin C,
- Stahlberg A,
- Aprico K,
- Larsson K,
- Yabe T,
- Moons L,
- Fotheringham A,
- Davies I,
- Carmeliet P,
- Schwartz JP,
- Pekna M,
- Kubista M,
- Blomstrand F,
- Maragakis N,
- Nilsson M,
- Pekny M