Poly-ICLC preconditioning protects the blood–brain barrier against ischemic injury in vitro through type I interferon signaling

Authors


Address correspondence and reprint requests to Mary P. Stenzel-Poore, PhD, Department of Molecular Microbiology and Immunology, L220, Oregon Health & Science University, 3181 Sam Jackson Park Road, Portland, OR 97239, USA. E-mail: poorem@ohsu.edu

Abstract

Preconditioning with a low dose of harmful stimulus prior to injury induces tolerance to a subsequent ischemic challenge resulting in neuroprotection against stroke. Experimental models of preconditioning primarily focus on neurons as the cellular target of cerebral protection, while less attention has been paid to the cerebrovascular compartment, whose role in the pathogenesis of ischemic brain injury is crucial. We have shown that preconditioning with polyinosinic polycytidylic acid (poly-ICLC) protects against cerebral ischemic damage. To delineate the mechanism of poly-ICLC protection, we investigated whether poly-ICLC preconditioning preserves the function of the blood–brain barrier (BBB) in response to ischemic injury. Using an in vitro BBB model, we found that poly-ICLC treatment prior to exposure to oxygen-glucose deprivation maintained the paracellular and transcellular transport across the endothelium and attenuated the drop in transendothelial electric resistance. We found that poly-ICLC treatment induced interferon (IFN) β mRNA expression in astrocytes and microglia and that type I IFN signaling in brain microvascular endothelial cells was required for protection. Importantly, this implicates a potential mechanism underlying neuroprotection in our in vivo experimental stroke model, where type I IFN signaling is required for poly-ICLC-induced neuroprotection against ischemic injury. In conclusion, we are the first to show that preconditioning with poly-ICLC attenuates ischemia-induced BBB dysfunction. This mechanism is likely an important feature of poly-ICLC-mediated neuroprotection and highlights the therapeutic potential of targeting BBB signaling pathways to protect the brain against stroke.

Abbreviations used
BBB

blood–brain barrier

BMECs

brain microvascular endothelial cells

CPT-cAMP

8-(4-Chlorophenylthio)adenosine 3′,5′-cyclic monophosphate sodium salt

CTR

control

DMEM

Dulbecco modified Eagle medium

EAE

autoimmune encephalomyelitis.

FBS

fetal bovine serum

ICAM-1

intercellular adhesion molecule–1

IFNAR

type I IFN receptor

IFN

interferon

MCAO

middle cerebral artery occlusion

MMPs

matrix metalloproteinases

OGD

oxygen-glucose deprivation

PBS

phosphate buffered saline

Pe

endothelial permeability coefficient

poly-ICLC

polyinosinic polycytidylic acid

RO 201724

4-(3-Butoxy-4-methoxybenzyl)imidazolidin-2-one

TEER

transendothelial electric resistance

TJ

tight junction

VCAM-1

vascular cell adhesion molecule–1

ZO-1

Zonula Occludens 1

Brain ischemia-reperfusion injury increases perivascular inflammation and blood–brain barrier (BBB) permeability, contributing to tissue damage. Endothelial cells of the cerebral microvasculature are the main component of the BBB. These cells are closely associated with astrocytes and microglia resulting in a complex network of cellular interactions that are essential for the maintenance of the nervous system microenvironment (Abbott et al. 2006). The loss of this interaction significantly contributes to the pathogenesis of ischemic brain injury (Endres et al. 2008). This has led to the concept that stroke injury is primarily a cerebrovascular disorder (Sandoval and Witt 2008; Weiss et al. 2009). The brain's resistance to ischemic injury can be augmented by preconditioning, which is defined as a brief exposure to a modest dose of a harmful stimulus that subsequently provides robust protection against a more severe insult (Dirnagl et al. 2009; Marsh et al. 2009). The majority of experimental preconditioning models focus on neurons as the cellular target of cerebral protection, while little attention has been paid to the cerebrovascular compartment. It has been shown that preconditioning stimuli can attenuate BBB disruption both in vitro (An and Xue 2009; Gesuete et al. 2011) and in vivo (Masada et al. 2001; Hua et al. 2008); however, the signaling pathways involved in the preconditioning-induced BBB protection remain unclear.

We have previously shown that preconditioning with a stabilized form of polyinosinic polycytidylic acid (poly-IC), a synthetic dsRNA and an innate immune activator, protects against cerebral ischemic damage (Packard et al. 2012). Poly-ICLC is a version of poly-IC stabilized with poly-l-lysine and carboxymethylcellulose that has shown clinical promise in humans for various indications (e.g., vaccines, multiple sclerosis, cancer, and viral infections) (Markosian and Hyde 2005; Rosenfeld et al. 2010). Poly-ICLC is a potent inducer of interferon (IFN)β, which is known to stabilize the BBB and reduce cellular infiltration into damaged brain regions following stroke (Veldhuis et al. 2003a, b; Kraus et al. 2004). Thus, we hypothesize that poly-ICLC preserves the function of the BBB in the setting of ischemic injury, potentially through the induction of IFNβ. Here, we investigated whether poly-ICLC preconditioning maintains the integrity of the BBB in modeled ischemia using an in vitro BBB model consisting of a co-culture of primary murine brain microvessel endothelial cells (BMEC) and primary mixed astrocytes and microglia cells. We tested whether protection depends on induction of IFNβ and consequent activation of type I IFN signaling. Our data indicate that preconditioning with poly-ICLC protects the BBB in the setting of modeled ischemia and suggests a potential mechanism of neuroprotection for this agent when given in advance of stroke in vivo. Such strategies that target these BBB signaling pathways may serve as new therapeutic tools for neuroprotection in stroke patients.

Methods

Mice

C57Bl/6J (WT) mice were obtained from Jackson Laboratories (West Sacramento, CA, USA). Type I IFN receptor deficient (IFNAR−/−) mice were provided by Dr. Herbert Virgin and Dr. Anthony French (Washington University School of Medicine, St. Louis, MO, USA). IFNβ−/− mice were provided by Dr. Tomas Leanderson (Lund University). All in vivo studies were performed with male mice between 8–12 weeks of age. Primary cultures were prepared from 1 to 2 day postnatal mice (mixed glial) or 8–12 week-old male mice (endothelial). All mice were given free access to food and water and housed in a facility approved by the Association for Assessment and Accreditation of Laboratory Animal Care International. Animal protocols were approved by the Oregon Health & Science University Institutional Animal Care and Use Committee (OWLAW# A3304-01) and met the guidelines set forth by the National Institutes of Health.

Drug treatments

For in vivo experiments, mice were given a subcutaneous injection of poly-ICLC (also known as Hiltonol®, which is the poly-IC stabilized with poly-l-lysine and carboxymethyl cellulose; 1.6 mg/kg, Oncovir, Washington DC, USA) or carboxymethyl cellulose vehicle (Sigma Aldrich, St. Louis, MO, USA) in a total volume of 100 μL. Mice were treated 72 h prior to the induction of ischemia. For in vitro experiments both compartments of the co-cultures were treated with vehicle or poly-ICLC (2 μg/mL) 24 h prior to oxygen-glucose deprivation (OGD).

BBB in vitro model

Primary cultures of mixed glial cells, containing both astrocytes and microglia, were prepared from 1 to 2 day postnatal mice (Kis et al. 2001). Brains were collected, meninges removed, and cortical pieces were mechanically dissociated and incubated for 15 min at 37°C with trypsin (2.5 mg/mL; Invitrogen, Grand Island, NY, USA) and DNAse I (0.08 mg/mL; Sigma). Supernatant containing the astroglial cells was collected, centrifuged, and resuspended in Dulbecco's modified Eagle medium (DMEM, Invitrogen) containing 0.5 mg/mL gentamicin (Sigma) and 10% fetal bovine serum (FBS; HyClone, Austin, TX, USA). Cells were seeded on poly-l-lysine (Sigma) coated 12-well plates. Glial cells were cultured for 3 weeks before use in experiments.

Brain microvessel endothelial cells (BMECs) were prepared from male mice (8–12 week-old) (Deli et al. 2003). Forebrains were collected in ice-cold sterile phosphate buffered saline (PBS), and the meninges were removed. Subsequently, the gray matter was minced to 1-mm3 pieces and digested with 1 mg/mL collagenase CLS2 (Worthington, Lakewood, NJ, USA) in DMEM for 45 min at 37°C. Microvessels were separated from myelin containing elements by centrifugation in 20% bovine serum albumin (Sigma)–DMEM (1000 g, 20 min), and further digested with 1 mg/mL collagenase–dispase (Roche, Mannheim, Germany) in DMEM for 30 min to remove pericytes from the basal membrane. Microvascular endothelial cell clusters were separated on a 33% continuous Percoll (Amersham, Piscataway, NJ, USA) gradient, collected, and seeded on collagen type IV and fibronectin coated cell culture inserts (Transwell clear, 1.12 cm2; pore size, 0.4 μm, Costar, Lowell, MA, USA). Cultures were maintained in DMEM supplemented with 5 μg/mL gentamicin (Sigma), 20% plasma-derived bovine serum (First Link, Wolverhampton, UK), 200 μg/mL endothelial cell growth supplement (Sigma), and 100 μg/mL heparin (Sigma). In the first 2 days, culture medium contained 4 μg/mL puromycin (Sigma) to selectively remove P-glycoprotein negative contaminating cells (Perriere et al. 2005). Cultures reached confluency within 4–5 days and then were used for experiments.

To induce typical BBB characteristics, BMECs were co-cultured with mouse cerebral astrocytes and microglia cells (Deli et al. 2003, 2005). Two days after seeding, the transwell inserts containing the BMECs were placed into multiwell plates containing the glial cells at the bottom, and the culture medium was replaced with fresh endothelial culture medium. Twenty-four hours later, culture medium was supplemented with 8-(4-Chlorophenylthio)adenosine 3′,5′-cyclic monophosphate sodium salt (CPT-cAMP) 250 μM (Sigma) and 4-(3-Butoxy-4-methoxybenzyl)imidazolidin-2-one (RO 201724) 17.5 μM (Sigma) for 24 h to tighten junctions and elevate resistance (Rubin et al. 1991; Deli et al. 2005; Perriere et al. 2005).

Measurement of monolayer resistance and permeability

Trans-endothelial electrical resistance (TEER), representing the permeability of TJs for sodium ions, was measured by an EVOM resistance meter (World Precision Instruments, Sarasota, FL, USA) using STX-2 electrodes and was expressed relative to the surface area of endothelial monolayer (Ωcm2). The TEER of cell-free inserts (80–90 Ωcm2) was subtracted from the values.

For permeability, measurements of the flux of Na-fluorescein (Sigma), as an index of paracellular transport, and of peroxidase-conjugated albumin (Jackson ImmunoResearch, West Grove, PA, USA) as an index of transcelluar pathway, across endothelial monolayers were determined as previously described (Kis et al. 2001). Briefly, cell culture inserts with BMECs were transferred to 12-well plates containing 1.5 mL/well of DMEM without phenol red. In upper chambers, culture medium was replaced by 500 μL DMEM without phenol red containing 10 μg/mL Na-fluorescein (MW: 376 Da) and 0.5 μg/mL of peroxide albumin (MW: 67 kDa). After 15, 30, 60, 120, and 240 min, the inserts were transferred to a new well containing DMEM without phenol red. The concentrations of the marker molecules were determined in samples obtained from the lower compartments. Albumin and Na-fluorescein concentrations were measured by a plate reader (Spectra Max 190, absorbency 450 nm for Albumin and Spectra Max Gemini XS emission: 515 nm, excitation: 460 nm for Na-fluorescein, both from Molecular Devices, Sunnyvale, CA, USA). Flux across cell-free inserts was also measured. Transport was expressed as ‘μL’ of donor (upper) compartment volume from which the tracer is completely cleared (Deli et al. 2005):

cleared volume (μL) = concentrationlower × volumelower × concentration−1Upper

The average cleared volume was plotted versus time, and permeability × surface area product value for endothelial monolayer (PSe) was calculated by the following formula:

PS−1endothelial = PS−1total−PS−1insert

PSe divided by the surface area generated the endothelial permeability coefficient (Pe, in 10−3 cm/min).

Oxygen-glucose deprivation (OGD)

Co-cultures were incubated in an anaerobic atmosphere of 85% N2, 10% CO2, and 5% H2 at 37°C for 5 h. Once in the hypoxic chamber, the culture medium was replaced with deoxygenated glucose-free medium. The anaerobic conditions within the chamber were monitored using an electronic oxygen/hydrogen analyzer (Coy Laboratories, Grass Lake, MI, USA). After the OGD period, medium was replaced with normoglycemic medium, and the cells were returned to a normoxic incubator (reoxygenation) for up to 24 h. In control co-cultures, the medium was replaced with fresh complete culture medium and the cells were not exposed to OGD.

Immunofluorescence

BMECs were fixed with ice-cold ethanol for 30 min at 4°C. After 30 min incubation with 0.1% triton for permeabilization followed by 30 min with FBS 3% for blocking, the cells were incubated with the primary antibodies rabbit anti-ZO-1 (1 : 100; Zymed, Grand Island, NY, USA) and rabbit anti-occludin (1 : 100; Zymed) overnight at 4°C. Anti-rabbit secondary antibody (1 : 100, Chemicon, Billerica, MA, USA) was used. Immunofluorescent staining was acquired by an Olympus BX61 microscope equipped with an Olympus confocal scan unit FV1000 (Olympus, Hicksville, NY, USA).

IFNα and IFNβ ELISA

Culture media was collected at 6, 8, and 24 h after poly-ICLC treatment and 24 h after OGD. The media was immediately concentrated using centrifugal filters (Amicon Ultra, 3K, Millipore, MA, USA) for 30 min at 14 000 g and then stored at −80°C until analysis. The IFNα and IFNβ levels in the culture media were measured by ELISA kits (PBL Interferon Source, Piscataway, NJ, USA) according to manufacturer's instructions.

RNA isolation, reverse transcription, and quantitative PCR

RNA was isolated from both endothelial and glial cells from individual wells at 6, 8, and 24 h after poly-ICLC treatment and 24 h after OGD using a Mini RNA isolation kit (Qiagen, Valencia, CA, USA). Reverse transcription was performed on 1 μg of RNA using Omniscript (Qiagen). Quantitative PCR was performed using Taqman Gene Expression Assays (Applied Biosystems, Carlsbad, CA, USA) for each gene of interest on StepOne Plus machine (Applied Biosystems). Results were normalized to β-Actin expression and analyzed relative to their control counterparts. The relative quantification of the gene of interest was determined using the comparative CT method (2-DDCt).

Mouse ischemia-reperfusion model

Focal cerebral ischemia was induced by middle cerebral artery occlusion (MCAO) as described previously (Stevens et al. 2008). In brief, MCAO was performed in anesthetized mice (1.5–2% isoflurane) by threading a 7-0 silicon-coated nylon surgical filament (Doccol, Redlands, CA, USA) through the external carotid artery to the internal carotid artery, blocking blood flow at the bifurcation of the MCA and anterior cerebral artery for 45 min. Following occlusion, the filament was removed and blood flow was restored. Cerebral blood flow was monitored throughout the procedure by laser Doppler flowmetry (Transonic System Inc., Ithaca, NY, USA), and animals were excluded if blood flow was not reduced by 80% or greater during occlusion. Body temperature was maintained at 37°C during the surgery.

Evaluation of infarct size

Twenty-four hours following MCAO, mice were deeply anesthetized with isoflurane and then perfused with ice-cold saline containing 2 U/mL sodium heparin. Brains with olfactory bulbs removed were sectioned into 1-mm slices beginning from the rostral end, for a total of seven slices. The infarct area was visualized by incubating the sections in 1.5% 2,3,5-triphenyltetrazolium chloride (TTC; Sigma) in PBS for 10 min at 37°C (Sigma Aldrich). Sections were then imaged and the infarct area was measured using ImageJ software (NIH Image, Bethesda, MD, USA). Infarct volume was calculated using the indirect method [(contralateral live − ipsilateral live)/contralateral live × 100] to account for the effects of edema, and the final infarct data are given as% damage.

Statistical analysis

Data are represented as mean ± SEM. All in vitro experiments were performed with four wells per time point within an experiment, and experiments were repeated two or more times. Statistical analysis was performed on combined experiments using GraphPad Prism5 software (La Jolla, CA, USA). Two-way anova with Bonferroni post hoc test was used for TEER and in vivo studies, one-way anova with Bonferroni post hoc test for Pe, ELISA and mRNA analysis.

Results

Poly-ICLC preconditioning significantly attenuated OGD-induced effects on the BBB

The in vitro BBB model used for the experiments consisted of BMECs co-cultured with primary mixed astrocytes and microglia cells. Here, we evaluated the effect of poly-ICLC preconditioning on the in vitro BBB subjected to OGD (Fig. 1a). Poly-ICLC (10 ng/mL to 2 μg/mL) was administered 24 h prior to 5 h OGD. BMECs trans-endothelial electrical resistance (TEER) and endothelial permeability coefficient (Pe) for Na fluorescein and Albumin were evaluated at 24 h post OGD (Fig. 1a). After four days of co-culture, control BMECs showed high TEER (218.8 ± 6.077 Ωcm2) and low Pe values for Na fluorescein (0.67 ± 0.062×10−3 cm/min) and Albumin (0.022 ± 0.0049×10−3 cm/min) (CTR, Fig. 1b and c). These values are typical for the murine BBB model and considered representative of a functional BBB (Deli et al. 2005; Gesuete et al. 2011). OGD caused a drastic reduction of TEER value (81.39 ± 3.5 Ωcm2, Fig. 1b) and a significant increase in endothelial permeability coefficient (Pe) for both Na fluorescein (1.04 ± 0.12×10−3 cm/min) and Albumin (0.68 ± 0.16×10−3 cm/min), Fig. 1c compared with control co-cultures at 24 h after OGD. Co-cultures pre-treated with 2 ug/mL poly-ICLC showed TEER values significantly higher (137.9 ± 7.7 Ωcm2, Fig. 1b) and Pe values for Na fluorescein significantly lower (0.52 ± 0.026×10−3 cm/min, Fig 1c) compared with the OGD group. All the poly-ICLC doses tested significantly lowered the Pe values for Albumin compared with the OGD group (Fig 1c). These data showed that poly-ICLC preconditioning protects the BBB by attenuating the OGD-induced effect on the TEER and maintaining basal levels of permeability in response to OGD. Poly-ICLC at 2 ug/mL showed the best protective effect on all the outcomes measured, accordingly, the 2 ug/mL dose was used for all subsequent experiments.

Figure 1.

Polyinosinic polycytidylic acid (Poly-ICLC) preconditioning maintained in vitro blood-brain-barrier (BBB) features following oxygen-glucose deprivation (OGD). (a) Experimental design: brain microvascular endothelial cells (BMECs) and glial cells were co-cultured for 2 days followed by preconditioning with poly-ICLC (10 ng/mL to 2 μg/mL). Twenty-four hours later co-cultures were exposed to 5 h of OGD. Poly-ICLC preconditioning at 2 ug/mL significantly prevented the drastic reduction of transendothelial electric resistance (TEER) values (b) and the increase of endothelial permeability coefficient (Pe) values for Na fluorescein and Albumin (c) induced by OGD. Data are reported as mean ± SEM; n = 3 individually repeated experiments. **p < 0.01 and ***p < 0.001 versus CTR; °p < 0.05 and °°p < 0.01 versus OGD, one-way anova followed by Bonferroni test.

Poly-ICLC treatment resulted in a selective robust induction of IFNβ

Poly-ICLC treatment is associated with robust production of both IFNα and IFNβ (Levy et al. 1975; Levy and Levine 1981; Bever et al. 1985, 1988; Longhi et al. 2009). To evaluate the effect of poly-ICLC on type I IFN induction in the in vitro BBB system, we measured both IFNα and IFNβ in the cell culture medium collected at 6, 8, and 24 h after poly-ICLC treatment and 24 h after OGD. We found that poly-ICLC caused a significant increase of IFNβ protein levels at 6 and 8 h after the treatment (151 ± 33.5 pg/mL and 178 ± 14.73 pg/mL, respectively) compared with control cells (1.01 ± 1.25 pg/mL, Fig. 2a), while no increase in IFNβ at 24 h after poly-ICLC treatment or 24 h following OGD was detected. IFNα was not detected at any time point following poly-ICLC treatment (data not shown), thus Poly-ICLC selectively induced IFNβ in our model.

Figure 2.

Polyinosinic polycytidylic acid (Poly-ICLC) treatment affected IFNβ induction. (a) IFNβ protein levels were measured by ELISA in culture medium collected at 6, 8, and 24 h after poly-ICLC treatment and 24 h post oxygen-glucose deprivation (OGD). Poly-ICLC resulted in induction of IFNβ protein levels at 6 and 8 h after the treatment. (b) IFNβ mRNA was evaluated in mixed glial cells at 6, 8, and 24 h after poly-ICLC treatment and 24 h post OGD. IFNβ was induced at all the time points tested compared with control vehicle treated cells. Data are reported as mean ± SEM; n = 2 individually repeated experiments. One-way anova followed by Bonferroni test, ***p < 0.001 versus CTR.

To understand what cellular population was responsible for the IFNβ production, we evaluated IFNβ mRNA expression in both endothelial and glial cells. We evaluated IFNβ mRNA at 6, 8, and 24 h after poly-ICLC treatment and 24 h after OGD (Fig 2b). At all the time points tested, we found that IFNβ mRNA was significantly induced in glial cells treated with poly-ICLC compared with control cells (Fig 2b). In contrast, endothelial cells showed no induction of IFNβ mRNA expression (data not shown).

Interferon-β produced by astrocytes and microglia cells was required for poly-ICLC-induced BBB protection

To evaluate whether the induction of IFNβ by the astrocytes and microglia cells was crucial for the poly-ICLC preconditioning, we used IFNβ−/− mice. The BBB system was modeled with WT BMECs in co-culture with glial cells isolated from IFNβ−/− mice (Fig 3). We found that poly-ICLC preconditioning did not protect the BBB against OGD, when IFNβ−/− glial cells were used in the co-culture. As shown in Fig. 3, both TEER (Fig 3a) and the Pe values for Na fluorescein and Albumin in the poly-ICLC treated group (Fig 3b) were similar to the vehicle-treated OGD group. These data indicate that the IFNβ produced by astrocytes and microglial cells played a key role in the poly-ICLC-induced BBB protection.

Figure 3.

IFNβ produced by astrocytes and glial cells was required for polyinosinic polycytidylic acid (Poly-ICLC) preconditioning. Poly-ICLC preconditioning was tested in the in vitro blood–brain barrier (BBB) modeled with WT brain microvascular endothelial cells (BMECs) in co-culture with astrocytes and microglia isolated from IFNβ−/− mice. These poly-ICLC treated co-cultures were not protected against oxygen-glucose deprivation (OGD) as demonstrated by transendothelial electric resistance (TEER) (a) and Pe values for both Na fluorescein and Albumin (b) comparable to the vehicle-treated OGD group. Data are reported as mean ± SEM; n = 2 individually repeated experiments. **p < 0.01 and ***p < 0.001 versus CTR; two-way anova followed by Bonferroni test for TEER and one-way anova followed by Bonferroni test for Pe.

Type I IFN signaling on endothelial cells was required for poly-ICLC-induced protection

To determine whether type I IFN signaling was required for poly-ICLC preconditioning and which cell types utilized type I IFN signaling, we set up the following in vitro BBB models: WT BMECs in co-culture with type I IFN receptor deficient (IFNAR−/−) mixed glial cells (Fig. 4a) or IFNAR−/− BMECs in co-culture with WT mixed glial cells (Fig. 4b). Poly-ICLC preconditioning was still effective, when type I IFN signaling was absent on astrocytes and microglial cells (Fig. 4a), as shown by the significant attenuation of TEER values (24 h post OGD: 102.5 ± 3.9 Ωcm2) and Pe for Na fluorescein (0.45 ± 0.033×10−3 cm/min) and Albumin (0.006 ± 0.001×10−3 cm/min) in the poly-ICLC treated group compared with OGD group (TEER 24 h post OGD: 73.8 ± 5.9 Ωcm2; Pe for Na fluorescein: 0.7 ± 0.025×10−3 cm/min; Pe for Albumin: 0.03 ± 0.002×10−3 cm/min). Alternatively, when type I IFN signaling was absent on endothelial cells, poly-ICLC preconditioning was no longer effective in inducing protection (Fig. 4b), suggesting a dominant role of endothelial-mediated type I IFN signaling in poly-ICLC preconditioning.

Figure 4.

Polyinosinic polycytidylic acid (Poly-ICLC) preconditioning required type I IFN signaling on endothelial cells. Poly-ICLC preconditioning was tested in the in vitro blood–brain barrier (BBB) modeled with WT brain microvascular endothelial cells (BMECs) in co-culture with type I IFN receptor knockout (IFNAR−/−) glial cells (a) or IFNAR−/− BMECs in co-culture with WT glial cells (b). Poly-ICLC protected the barrier integrity when type I IFN signaling was absent in glial cells, as shown by attenuated transendothelial electric resistance (TEER) reduction and lower Pe values for Na fluorescein and Albumin in poly-ICLC group compared with oxygen-glucose deprivation (OGD) (a). Instead, poly-ICLC preconditioning was not effective when type I IFN signaling was absent on the BMECs, in this case, poly-ICLC and OGD group showed similar TEER and Pe values (b). Data are reported as mean ± SEM; n = 3 individually repeated experiments. *p < 0.05, **p < 0.01 and ***p < 0.001 versus CTR; °p < 0.05, °°p < 0.01 and °°°p < 0.001 versus OGD, two-way anova followed by Bonferroni test for TEER and one-way anova followed by Bonferroni test for Pe.

Poly-ICLC attenuated OGD-induced Tight junction (TJ) loss in WT, but not in IFNAR-/- BMECs

TJ proteins play an essential role in maintaining the tightness of cell–cell interactions in the BBB endothelial cells. By immunofluorescence, we assessed the presence and distribution of Zonula Occludens 1 (ZO-1), a protein located in the cytoplasm and attached to the tail of transmembrane TJ proteins and of occludin, an integral membrane TJ protein, in WT and IFNAR−/− BMECs in our preconditioning model. Both WT and IFNAR−/− control BMECs showed an intense and continuous staining of ZO-1 and occludin at the cell borders (Figs 5 and 6, left panels). Exposure to 5 h of OGD resulted in a decreased intensity, loss of the continuous junctional staining pattern, and a global disorganization of the both ZO-1 and occludin protein (Figs 5 and 6, middle panels). In accordance to the TEER and permeability data previously shown, staining intensity and distribution of ZO-1 and occudin were maintained in WT BMECs pre-treated with poly-ICLC, while IFNAR−/− BMECs pre-treated with poly-ICLC showed a staining pattern similar to vehicle-treated OGD (Figs 5 and 6, right panels).

Figure 5.

Polyinosinic polycytidylic acid (Poly-ICLC) preconditioning prevented oxygen-glucose deprivation (OGD)-induced ZO-1 disorganization in WT but not in IFNAR−/− brain microvascular endothelial cells (BMECs). ZO-1 immunofluorescence showed an intense and continuous staining on the cell borders for both WT and IFNAR−/− control BMECs (left panels). OGD induced a decrease in immunostaining intensity, loss of the continuous junctional staining pattern, and a global disorganization of the proteins (middle panels). Poly-ICLC treated WT BMECs, but not poly-ICLC treated IFNAR−/− BMECs, showed a preserved staining intensity and distribution of ZO-1 following OGD. Scale bar = 10 μm.

Figure 6.

Polyinosinic polycytidylic acid (Poly-ICLC) preconditioning prevented oxygen-glucose deprivation (OGD)-induced occludin disorganization in WT but not in IFNAR−/− brain microvascular endothelial cells (BMECs). Occludin immunofluorescence showed an intense and continuous staining on the cell borders for both WT and IFNAR−/− control BMECs (left panels). Exposure to OGD decreased immunostaining intensity, reduced the continuous junctional staining pattern, and caused a global disorganization of the proteins (middle panels). Poly-ICLC treated WT BMECs, but not poly-ICLC treated IFNAR−/− BMECs, showed a preserved staining intensity and distribution of occludin following OGD. Scale bar = 10 μm.

Poly-ICLC preconditioning required type I IFN signaling in vivo

To determine whether type I IFN signaling was involved in poly-ICLC neuroprotection in an intact animal, WT and IFNAR−/− mice were treated with poly-ICLC or vehicle 72 h prior to MCAO (n = 7–10/group), (Fig. 7). Infarct volume measured at 24 h following MCAO revealed that WT mice were significantly protected by poly-ICLC preconditioning treatment, but IFNAR−/− mice were not protected (WT vehicle 34.7 ± 3.1 vs. poly-ICLC 13.8 ± 2.2; IFNAR−/− 31.5 ± 6.9 vs. poly-ICLC 35.4 ± 4.9 p > 0.05; Fig. 7). Therefore, these results support our in vitro data showing that type I IFN signaling is required for poly-ICLC mediated protection.

Figure 7.

Type-1 IFN receptor signaling was required for polyinosinic polycytidylic acid (Poly-ICLC) preconditioning in vivo. WT and IFNAR−/− mice were preconditioned with poly-ICLC or vehicle 72 h prior to MCAO (n = 7–10/group). Infarct size was determined 24 h following MCAO by TTC staining. WT mice were significantly protected by poly-ICLC preconditioning, while IFNAR−/− mice were not. Data are reported as means ± SEM; **p < 0.01 versus vehicle control by two-way anova and Bonferroni post hoc test.

Discussion

This study shows for the first time that preconditioning with poly-ICLC attenuates ischemia-induced BBB dysfunction. We found that the administration of poly-ICLC maintains both the function (prevention of the drastic reduction in TEER induced by OGD and restoration of Pe transport values) and integrity (preservation of TJ structure) of the BBB endothelial cells. In addition, our results point out a key role for IFNβ in the protective effect induced by poly-ICLC preconditioning on the BBB. We found that poly-ICLC induces astrocyte/microglial cell production of IFNβ prior to OGD, and we demonstrate that this induction is required for the protective effect. We further show that the IFNβ produced by the glial cells signals through type I IFN receptors on the endothelium to maintain the integrity of the BBB in response to OGD. The significance of these findings is demonstrated in vivo, where we show that poly-ICLC preconditioning depends on type I IFN signaling to protect the brain against ischemic injury.

The in vitro BBB model we used showed TEER values in the range of 160–210 Ωcm2. Although these values are far from the estimated in vivo BBB TEER (1500–6000 Ωcm2), findings in the present and previous studies (Deli et al. 2005) indicate that these cells retain an endothelial-specific phenotype, as assessed by the expression of markers (i.e., ZO-1 and occludin) as well as the ability to form a functionally tight barrier (i.e., low permeability values). In addition, available evidence supports the notion that the use of the mixed glial cells in the co-culture provide the endothelial cells with factors that help maintain their vascular phenotype (Cecchelli et al. 2007). All these features make our system a well-established in vitro BBB model retaining the features of a functional BBB (Deli et al. 2005).

The direct involvement of the BBB in the protective effect induced by preconditioning has been shown using brief ischemia (OGD) as the preconditioning stimulus in a similar in vitro BBB system (An and Xue 2009; Gesuete et al. 2011). Like poly-ICLC, stimulation with brief OGD 24 h prior to an extended episode of OGD is able to maintain BBB functions and integrity. An important feature that emerges comparing these two preconditioning models is the key role of astrocyte and microglial cells in mediating BBB protection. Gesuete et al. investigated the specific contribution of endothelial and mixed glial cells to the preconditioning-induced BBB protection by individually exposing the cell populations to the preconditioning stimulus. They found that preconditioned astrocytes/microglial cells were much more efficient in eliciting the protective phenotype than endothelial cells. Therefore, they suggest that molecules produced by preconditioned glial cells are able to induce BBB protection (Gesuete et al. 2011). In agreement with this finding, our data suggest that the protective BBB phenotype induced by poly-ICLC is because of its action on the glial cells, indicating that cross talk between glial and endothelial cells is important for establishing protection of the BBB.

Our data suggest that poly-ICLC mediates protection of the BBB through the induction of IFNβ. It is known that poly-ICLC treatment is associated with robust production of both IFNα and IFNβ (Levy et al. 1975; Levy and Levine 1981; Bever et al. 1985, 1988; Longhi et al. 2009). We found that poly-ICLC treatment resulted in induction of IFNβ by mixed glial cells; however, endothelial cells failed to induce either IFNα or IFNβ in response to the protective dose of poly-ICLC. It has previously been shown that IFNβ administered systemically at the time of or after stroke reduces infarct volume and improves blood–brain barrier integrity (Liu et al. 2002; Veldhuis et al. 2003a, b). A direct effect of IFNβ on brain endothelial cells has also been showed by Kraus et al. They found that human recombinant IFNβ consistently stabilizes both TEER and permeability of brain endothelial monolayers regardless of the experimental methods (monoculture vs. co-culture), the use of primary or immortalized bovine brain capillary endothelial cells, or species (Kraus et al. 2008).

In addition to stabilizing BBB permeability, IFNβ has also been shown to preserve BBB integrity through its effect on the TJ proteins (Kraus et al. 2004). We found that both ZO1 and occludin, components of the TJs, were maintained in BMECs treated with poly-ICLC compared with untreated cells subjected to OGD. Interestingly, we found that poly-ICLC was unable to prevent the breakdown of the ZO1 and occludin in IFNAR−/− BMECs. This suggests that IFNβ stimulates IFNAR on the endothelial cells and activates an intracellular pathway that preserves TJ proteins in the setting of OGD. A possible mechanism may be the inhibition of matrix metalloproteinases (MMPs). The MMPs are zinc- and calcium-dependent endopeptidases, identified as matrix-degrading enzymes and expressed in multiple cell types including glial and endothelial cells (Candelario-Jalil et al. 2009). MMPs cleave most components of the extracellular matrix including fibronectin, laminin, proteoglycans, and type IV collagen. By degrading neurovascular matrix and disruption of the extracellular tail of the BBB tight junctions, MMPs (mainly MMP-9) promote BBB damage, brain edema, and hemorrhage during acute ischemic stroke (Jin et al. 2010). IFNβ treatment inhibits MMP gene expression (Bauvois et al. 2002; Benveniste and Qin 2007) and MMP enzymatic activity in cellular systems including astroglioma cell lines and primary human and murine astrocytes (Ma et al. 2001). Thus, in our studies of poly-ICLC, the IFNβ derived from astrocytes and microglial cells may interact with IFNAR on the endothelial cells to inhibit endothelial MMPs and preserve TJ structures in response to OGD.

Additionally, IFNβ induced by poly-ICLC may also induce BBB protection by decreasing perivascular inflammation through the regulation of cell adhesions molecules such as intercellular adhesion molecule–1 (ICAM-1) and vascular cell adhesion molecule–1 (VCAM-1). Ischemia/reperfusion leads to up-regulation of these cell adhesion molecules that results in increased leukocytes transmigration (Huang et al. 2006). The activation of the adhesion molecules also results in the induction of different signaling pathways such as increase in intracellular Ca2+ concentrations, increase of reactive oxygen species, and gene transcription activation (Wang and Doerschuk 2002). Additionally, activation of adhesion molecules can induce stress fiber formation, cellular shape changes, and cytoskeleton rearrangement that can play a key role in BBB permeability (Mousa 2008). In vitro treatment of brain endothelial cells with IFNβ decreases the abundance of both ICAM-1 and VCAM-1 (Benveniste and Qin 2007). The authors speculated that IFNβ exerts direct anti-inflammatory effects on brain endothelial cells by inhibiting the production of adhesion molecules, which results in the decreased entry of leukocytes into the CNS (Benveniste and Qin 2007). In addition, An et al. showed that OGD/reoxygenation increased the expression of ICAM-1 and VCAM-1 in rat BMECs and that OGD preconditioning alleviated these changes, correlating decreased inflammation with the protective effect (An and Xue 2009). Modulation of the MMPs and adhesion molecules both likely contribute to the protection induced by the IFNβ produced by the glial cells after the stimulation with poly-ICLC.

We have also shown that the presence of type I IFN receptor is crucial for the poly-ICLC preconditioning both in vitro and in vivo. Type I IFN signaling has been shown to have an important role in different CNS pathologies. In a mouse model of autoimmune encephalomyelitis (EAE), an experimental model of multiple sclerosis, the absence of IFNAR is associated with exacerbated clinical disease accompanied by a markedly higher inflammation, demyelination, and increased numbers of infiltrating macrophages cells in the CNS (Prinz et al. 2008). Khorooshi et al. also showed that IFNAR-deficient mice subjected to transection of entorhinal afferents, an experimental model of CNS sterile injury, have increased leukocyte infiltration in the lesion-reactive hippocampi (Khorooshi and Owens 2010). Interestingly, they also found that IFNAR-deficient mice showed increased MMP9 mRNA in the lesion side, suggesting that the type I IFN signaling is important in regulating the mechanisms involved in BBB disruption (Khorooshi and Owens 2010).

Our findings demonstrate the poly-ICLC induces production of IFNβ by glial cells, which then binds to and signals IFNARs on endothelial cells. Although not evident in the literature, our data suggest that IFNAR is expressed on the abluminal side of the BBB placing it in close proximity to astrocytes and microglia cells allowing a paracrine effect of the cytokine. IFNβ signaling of the endothelium results in subsequent suppression of MMPs and cell adhesion molecules (VCAM-1, ICAM-1) both of which are associated with ischemic disruption of the BBB (Fig. 8).

Figure 8.

Schematic model of polyinosinic polycytidylic acid (Poly-ICLC) mechanism of action on the blood–brain barrier (BBB). Ischemic injury causes BBB damage by increasing matrix metalloproteinases (MMPs) activity that results in degradation of TJs proteins and consequent increased permeability. In addition, ischemic injury increases the expression of the adhesion molecules on the endothelial surface resulting in augmentation of leukocyte transmigration (left). Poly-ICLC preconditioning protects BBB endothelial cells against ischemic injury (right). The possible poly-ICLC-induced mechanism of protection involves IFNβ induction by astrocytes and microglia cells. The IFNβ can then interact with type I IFN receptor present on the abluminal side of endothelium. The consequent activation of the endothelial type I IFN signaling pathways confers resistance against ischemic injury-mediated disruption of barrier function, possibly through preservation of the TJ proteins and prevention of the adhesion molecules expression.

These studies add to our fundamental understanding of the neuroprotective effect of poly-ICLC preconditioning. Here, we show that the mechanism of action of poly-ICLC is mediated, in part, through the effect of IFNβ on the BBB. Collectively, these studies offer further promise for poly-ICLC as an antecedent therapeutic approach against ischemic injury.

Acknowledgements

Poly-ICLC was kindly provided by Dr. Andres Salazar, MD, Oncovir, Washington DC. This work was supported by funding from the National Institute of Neurological Disorders and Stroke NS050567 and NS062381.

Authorship credit

RG, SLS, KBV, AEBP, and MPSP conceived and designed the experiments; RG, AEBP, TY and VKC performed the experiments; RG, AEBP, and SLS analyzed the data; RG, SLS, KBV and AEBP wrote the paper; FRB and MSPS critically revised the article and gave conceptual advice; final approval of the version to be published was obtained from all authors.

Conflicts of interest

There are no conflicts of interest related to this study.

Ancillary