Targeting gliovascular connexins prevents inflammatory blood-brain barrier leakage and astrogliosis

The blood-brain barrier is formed by capillary endothelial cells expressing connexin 37 (Cx37), Cx40, and Cx43 and is joined by closely apposed astrocytes expressing Cx43 and Cx30. We investigated whether connexin-targeting peptides could limit barrier leakage triggered by LPS-induced systemic inflammation in mice. Intraperitoneal LPS administration increased endothelial and astrocytic Cx43 expression; elevated TNF-α, IL-1β, IFN-γ, and IL-6 in plasma and IL-6 in the brain; and induced barrier leakage recorded over 24 hours. Barrier leakage was largely prevented by global Cx43 knockdown and Cx43/Cx30 double knockout in astrocytes, slightly diminished by endothelial Cx43 knockout, and not protected by global Cx30 knockout. Intravenous administration of Gap27 or Tat-Gap19 peptides just before LPS also prevented barrier leakage, and intravenously administered BAPTA-AM to chelate intracellular calcium was equally effective. Patch-clamp experiments demonstrated LPS-induced Cx43 hemichannel opening in endothelial cells, which was suppressed by Gap27, Gap19, and BAPTA. LPS additionally triggered astrogliosis that was prevented by intravenous Tat-Gap19 or BAPTA-AM. Cortically applied Tat-Gap19 or BAPTA-AM to primarily target astrocytes also strongly diminished barrier leakage. In vivo dye uptake and in vitro patch-clamp showed Cx43 hemichannel opening in astrocytes that was induced by IL-6 in a calcium-dependent manner. We conclude that targeting endothelial and astrocytic connexins is a powerful approach to limit barrier failure and astrogliosis.


Animal dose response curve for LPS treatment
A dose response experiment with concentrations ranging between 1 mg/kg and 50 mg/kg was performed to estimate the optimal dose for further experiments, which was 25 mg/kg (corresponding to ~300 µg/mL blood). This dose leads to clear sickness behavior including decreased motor activity, withdrawal, and reduced food and water intake (Dantzer et al., 2008). As inflammation progressed, animals displayed lack of grooming (fuzzy fur), diarrhea and bleared eyes. Mice were anesthetized with a mixture of ketamine (20 mg/mL) and xylazine (0.1 %). The animals were intracardially injected with 10 µL heparin (2000 U/mL) and transcardially perfused with PBS. Immediately after isolation, the brains were snap-frozen in liquid nitrogencooled isopentane and stored at -80 °C. Fifty micron thick coronal brain sections were cut using a using a cryostat (Leica Biosystems) and mounted in Vectashield antifade mounting medium containing Dapi stain (Vector Labs). Banks et al. previously indicated that the brain shows regional variability in terms of LPS-induced BBB permeability increases with cortical area's showing a 20-55 % increase (Banks et al., 2015). The area from which sections were made was located between 0.94 mm and 2.5 mm posterior to the bregma, where the cortical region corresponds to the primary somatosensory cortex. Parenchymal fluorescence in the brain sections was visualized using a BD Pathway 435 BioImaging system (Becton Dickinson) equipped with a mercury light source that provides illumination from 360 nm to 700 nm, CCD camera and 10x objective. The signal intensity was determined using ImageJ software. For each section, fluorescence intensity was determined in 20 points, all located in the somatosensorial cortex. Background fluorescence was measured just outside the coronal sections and was subtracted from the fluorescence in the measurement points. To reduce variation among different experiments, results are expressed relative to control non-treated animals included in the same experiment.

Cranial window
Ketamine/Xylazine-anesthetized mice were fixed in a stereotactic frame. Using a dental drill, a craniotomy was made in the right parietal bone covering the somatosensorial cortex. The cranial window was 3 mm in diameter, centered 2 mm posterior to the bregma and 2 mm from the sagittal suture/midline. The dura was carefully removed, and the exposed cortex was loaded with artificial cerebrospinal fluid (aCSF; in mM: 1 CaCl2; 2 MgCl2; 126 NaCl; 1.25 Na 2 HPO 4 ; 2.5 KCl; 10 D-glucose and 25 Hepes) containing sulforhodamine 101 (SR101, 50 μM) further supplied with BAPTA-AM (2 mM) or Tat-Gap19 (200 µM) for interventions directed at intracellular Ca 2+ chelation or Cx43 hemichannel inhibition. After incubation of active compounds or vehicle, agarose (0.5% in aCSF) was used to seal the cranial window and the overlying skin was sutured. Subsequently, LPS was injected IP and animals were allowed to recover. At the indicated time points after LPS injection (3, 6 and 24 h), mice were anesthetized again for barrier permeability measurements as described above. Cortical fluorescence intensity was only measured in the SR101 loaded region.

Cortical hemichannel dye uptake experiments
LPS was injected IP and vehicle (aCSF) or Tat-Gap19 was applied to the exposed cortex 1 h prior to sacrification. After 30 min, ethidium bromide (EtBr, 100 µM) was added to the solution for the next 30 min. Next, mice were transcardially perfused with PBS, brains were collected in 4% PFA and subsequently snap-frozen in liquid nitrogen-cooled isopentane.
Twenty micron thick sections were cut and mounted in Dapi-containing Vectashield. EtBr uptake was visualized using a Leica SP8 confocal microscope.

Plasma and brain tissue preparation for luminex assay
Whole blood samples, obtained by cardiac puncture (prior to perfusion with PBS) and collected in heparinized test tubes, were centrifuged at 2000 g at 4 °C and for 15 min to deplete cells and platelets. The resulting supernatant (plasma) was carefully removed from the cell pellet and stored at -20°C. Brain samples were lysed in CHAPS buffer (3 mL/g tissue) containing CHAPS (0.5%), Hepes (10 mM), KCl (42 mM) and MgCl2 (5 mM). DTT (1 mM) and protease inhibitor cocktail (20 µL/mL) were freshly added prior to use. Protein concentration was determined using the Biorad DC protein assay kit (BioRad, Nazareth, Belgium) and absorbance was measured with a 590 nm long-pass filter.
Quantification of cytokines (TNFα, IL1β, IFNγ and IL6) in plasma and brain tissue lysates was performed using the BioPlex cytokine assay (BioRad), according to the instructions of the manufacturer. Circulating cytokine levels are expressed as pg/mL plasma, cerebral cytokine levels are expressed as pg/100µg total protein.

Cell isolation and cell culture studies
Isolation and culturing of primary mouse brain capillary endothelial cells was performed as described in (Coisne et al., 2005) with slight adaptations. Whole brains from decapitated C57BL/6 male mice (12 weeks) were collected in ice-cold phosphate buffered saline (PBS).
Meninges and large extracerebral vessels were removed using sterile lint and dry cotton swabs. Optic nerves, white matter and cerebellum were carefully dissected away. The remaining tissue was grounded in ice-cold Hank's Balanced Salt Solution (HBSS) containing 10 mM Hepes and 0.1 % BSA ('WBB'), using a 15 mL Dounce homogenizer (Wheaton, Fisher Scientific), first with loose and then tight pestle. The resulting homogenate was mixed with 30 % dextran solution and the suspension was centrifuged at 4 °C for 25 min at 3000 g. The resulting vascular pellet was resuspended and triturated in WBB after which it was filtered through a 60 µm nylon mesh that allows to discard larger sized vessels. The capillary-enriched filtrate was centrifuged at 1000 g for 7 min at room temperature (RT) and the vascular pellet was then digested in WBB containing 10 mg/mL collagenase-dispase (Roche Diagnostics), 1,47 µg/mL Tosyl Lysin Chloromethyl Ketone (Sigma-Aldrich) and 10 µg/mL DNase I (Roche Diagnostics), for 33 min at 37 °C in a shaking water bath. Enzyme digestion was stopped by adding excess WBB and the vascular pellet, obtained by centrifuging at 1000 g for 7 min was triturated one more time. A final centrifugation at 1000 g (7 min, room temperature) resulted in a vascular pellet that was resuspended in DMEM containing glucose (1 g/L), Na + -pyruvate (110 mg/L), glutamine (2 mM), gentamycin (50 µg/mL), BME vitamin solution (1%), BME amino acid solution (2%), basic fibroblast growth factor (bFGF, 1 ng/mL) and 20% newborn calf serum. The resulting cell suspension was plated onto matrigel-coated recipients and refreshed 24 h after seeding to remove red blood cells, cell debris and non-adherent cells.
For some experiments, we used the capillary-enriched filtrate (see images Fig. 2B and D).
Primary astrocytes were isolated from postnatal (P0-P1) cortices as previously described (Freitas-Andrade et al., 2019). Dissected cortices were triturated in DMEM and the resulting suspension was passed through a 70 μm cell filter before seeding into flasks. Cell culture medium (DMEM supplemented with 10% FBS, 10 units/mL penicillin, and 10 μg/mL streptomycin) was replaced 3 days after plating and every second day thereafter. After 7-8 days, astrocytes were harvested with trypsin-EDTA and frozen in freezing medium (90% FBS, and 10% DMSO). Frozen astrocytes were thawed and plated onto ploy-L-lysine (Sigma Aldrich) coated recipients. Cultures were maintained for 5-7 days prior to experiments.

Electrophysiological Recording
HeLa-Cx43 cells were seeded onto 13 mm diameter glass coverslips, RBE4 cells were seeded onto 13 mm diameter glass coverslips first coated with collagen and subsequently with Corning® Cell-Tak according to the supplier's instructions, primary astrocytes (55.000 cells) were plated onto 13 mm diameter glass coverslips coated with poly-L-lysine (Sigma Aldrich).  conductance histograms that displayed one or more Gaussian distributions. These were fit by a probability density function assuming independent channel opening (Ramanan and Brink, 1993;Wang et al., 2001;Wang et al., 2012;Wang et al., 2013). The samples were visualized by confocal microscopy (Leica TCS SP8 X; 63x water objective).
For evaluation of the connexin expression in BBB capillaries and perivascular endfeet, we used the ImageJ coloc plugin to count colocalized pixels. Counts were expressed relative to the number of CD31-positive endothelial cell pixels or AQP4-positive astrocytic endfeet pixels.
Immunohistochemical analysis of GFAP fluorescence was quantified from coronal brain cryosections and expressed relative to the signal observed in non-treated control animals (number of animals as in the treated group). Fluorescence was visualized by confocal microscopy (Leica TCS SP8 X; 63x water objective) and quantified in the somatosensory cortex.
Membranes were subsequently incubated with an alkaline phosphatase-conjugated goat anti-rabbit IgG antibody (Sigma-Aldrich) and detection was done using the nitro-bluetetrazolium/5-bromo-4-chloro-3-indolyl-phosphate reagent (NBT/BCIP kit, Zymed, Invitrogen). Alternatively, membranes were incubated with HRP-conjugated goat anti-rabbit IgG (Santa Cruz) and detection was done using SuperSignal TM West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific). Total protein stains by SYPRO® Ruby (Molecular probes) prior to antibody incubation, or detection with rabbit anti-β-tubulin antibody (Abcam) were used as loading controls. Quantification was done by drawing a rectangular window around the concerned protein band and determining the signal intensity using ImageJ. Background correction was done by the same procedure applied to nitrocellulose membranes where protein was absent. and Cx40 as determined in brain capillaries freshly isolated from mice. The expression levels tended to be decreased by LPS but this did not attain statistical significancy (unpaired t-test; n = 3). Antibodies were as follows: rabbit anti-Cx30 (1/500; Thermofisher) and chicken anti-AQP4 (1/1000; Synaptic systems); rabbit anti-Cx37 (1/500; Thermofisher), rabbit anti-Cx40 (1/200; Thermofisher) and rat anti-CD31 (1/100; Becton Dickinson).