Background
Endothelial cells play an essential role in normal homeostasis of the central nervous system (CNS). In healthy individuals, microvessels throughout most of the CNS possess a luminal monolayer of tightly apposed endothelial cells situated between the blood and brain parenchyma comprising together with adjacent astrocytes the blood-brain barrier (BBB) [
1]. Cerebral endothelial cells are crucial for normal neurological function as they constitute both a physical “barrier” which limits molecular and cellular exchange between blood and brain compartments and a “fence” which maintains polarity of transporters responsible for delivery of essential nutrients and removal of potentially harmful toxins [
2]. These CNS endothelia derive a low permeability barrier due to interendothelial tight junctions (TJ) occludin and claudin proteins as well as junction associated submembranous adaptor proteins such as zonula occludens (ZO)-1 [
3]. Several studies show membrane localization of tight junction proteins are the morphological correlate of BBB integrity and tightness [
4,
5]. Barrier disruption secondary to tight junction dysregulation results from reduced endovascular flow [
6], hypoxia/ischemia [
7], and inflammatory cytokines such as tumor necrosis factor-α (TNF-α) [
8] and vascular endothelial-derived growth factor (VEGF) [
9]. Several CNS diseases including neoplasia, hereditary vascular malformation, trauma, and chronic inflammatory and neurodegenerative diseases such as multiple sclerosis (MS) feature characteristics of BBB breakdown [
10,
11]. Characterizing factors able to influence BBB integrity and aspects of vascular remodeling during CNS inflammation may identify key molecules with both physiological and perhaps pathological roles in disease.
BBB cytoarchitecture and response to stimuli are often examined in a simplified treatment and effect system comprised of in vitro cultures of endothelial cells established from cerebral microvessels. These cells recapitulate in vivo BBB characteristics such as expression of specific endothelial markers (i.e., CD31 and VE-cadherin), BBB transporters (i.e., GLUT-1, P-glycoprotein, transferrin), and tight junction markers (i.e., ZO-1 and occludin) and form a monolayer with low paracellular permeability and high transendothelial electrical resistance (TEER) consistent with the presence of membrane-associated tight junctions [
12‐
14]. In the present study, we use a well-characterized in vitro model of the BBB consisting of immortalized human cerebral microvascular endothelial cells (hCMEC/D3) that express and appropriately localize important BBB proteins characteristic of their in vivo counterparts [
15,
16].
SPARC (secreted protein acidic and rich in cysteine) is a matricellular cell-matrix modulating protein involved in angiogenesis [
17,
18] and endothelial barrier function [
19]. Many cell types including endothelia, fibroblasts, and macrophages constitutively express SPARC and up-regulate its expression in tissue regions undergoing high rates of remodeling, repair, and proliferation [
20]. SPARC is typically enriched where new blood vessels are being formed, as evidenced using an in vivo chorioallantoic membrane (CAM) model of angiogenesis [
21]. In the CNS, SPARC is highly expressed in developing blood vessels at early stages of development and down-regulated with developmental maturity (Roskams Lab, unpublished observations). This spatiotemporal pattern of SPARC expression is consistent with the role for SPARC in angiogenesis and BBB establishment [
22,
23]. Normal physiological levels of SPARC in healthy individuals (0.1–0.8 μg/ml in plasma) are increased in neoplastic and inflammatory conditions (1.5–10 μg/ml in plasma) [
24‐
26]. Increased SPARC secretion has been associated with carcinoma [
27] and other tumors [
28] such as gliomas [
29], as well as inflammatory renal disease [
30,
31], and scleroderma characterized by vascular dysfunction, autoantibody production, and tissue fibrosis [
32].
SPARC may play a role in mediating the CNS response to injury and repair given enhanced expression in in vivo models of CNS damage and repair. In an in vivo cortical wound model, SPARC messenger RNA (mRNA) was abundantly expressed in developing blood vessels proximal to the wound edge days 3 to 10 post-injury, suggesting its involvement in the vascular response to CNS injury and cerebrovascular ischemia [
22,
33]. Furthermore, SPARC may be associated with neurological recovery following CNS injury. Proteomic screens of murine lamina propria olfactory ensheathing cell (LP-OEC)-conditioned media identified SPARC as the key secreted protein supporting neural tissue repair after damage, capable of promoting spinal cord repair by limiting gliotic scar and cavity formation, stimulating axonal outgrowth and directing angiogenesis [
34]. SPARC has been shown to promote Schwann cell-mediated neurite outgrowth in vivo and in vitro [
34]. Moreover, SPARC-null OECs transplanted into contused rat spinal cord reduced outgrowth of specific subsets of sensory and supraspinal axons and impaired immune response to injury, suggesting its role in neural regenerative processes and the neuroimmune response to CNS injury [
34].
Altogether, its potential roles in modulating angiogenic and barrier parameters of vascular development and repair [
19,
23,
35,
36], its expression and influence on neural regeneration after CNS injury [
33,
34], and its influence on the profile and extent of immune infiltration [
37,
38] make SPARC a molecule of particular interest in chronic neuroinflammatory pathologies such as MS.
Methods
Reagents and antibodies
Recombinant human (rh)-SPARC (R&D Systems, Minneapolis, MN), TNF-α (Invitrogen, Camarillo, CA), interferon gamma (IFN-γ), (Invitrogen, Camarillo, CA), and lipopolysaccharide (LPS) (Sigma-Aldrich, St. Louis, MO) were obtained commercially. Antibodies for immunoblotting and immunocytochemistry included monoclonal mouse anti-human SPARC IgG1 (2.5 μg/ml; Haematologic Technologies, Essex Junction, VT), polyclonal rabbit anti-ZO-1 IgG (1 μg/ml; Invitrogen), monoclonal mouse anti-ZO-1 IgG1 (2.0 μg/ml; Invitrogen), monoclonal mouse anti-occludin IgG1 (0.5 μg/ml; Zymed, Carlsbad, CA), monoclonal mouse anti-Ki-67 IgG1 (1:100; Millipore, Billerica, MA), and monoclonal mouse anti-human glyceraldehyde-3-phosphate dehydrogenase IgG1 (GAPDH, 1:10,000; Santa Cruz Biotechnology, Santa Cruz, CA). Secondary antibodies included Alexa Fluor 568 goat anti-mouse IgG (1:200; Invitrogen); Alexa Fluor 488 goat anti-rabbit IgG (1:200; Invitrogen); and horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (1:5000; Jackson ImmunoResearch, Burlington, ON). Isotype-matched control abs included mouse IgG1 (2.5 μg/ml; Invitrogen) and rabbit IgG (1 μg/ml; Invitrogen).
Cell culture
The hCMEC/D3 line was generously provided by Drs. B. Weksler, I. Romero, and P-O. Couraud (Cochin Institute, Paris). The cell line was established as described in Weksler et al. 2005. hCMEC/D3s were cultured in complete media comprised of EBM-2 media (Lonza, Walkersville, MD) supplemented with 5 % fetal bovine serum (FBS) (PAA Laboratories, Etobicoke, ON), 1 % penicillin-streptomycin, 1.4 μM hydrocortisone, 5 μg/ml L-ascorbic acid, 10 mM HEPES (all from Sigma), 1:100 chemically defined lipid concentrate (Invitrogen-Gibco), and 1 ng/ml basic fibroblast growth factor (bFGF) (Invitrogen-Gibco, Grand Island, NY) at 37 °C with 5 % CO2, 95 % air, and saturated humidity. The media was replenished every 2 to 3 days. For immunoblotting and reverse transcription polymerase chain reaction (RT-PCR), the cells were seeded at 1.2 × 10E4 cells/cm2 on 25 cm2 flasks and six-well plates (Corning, Corning, NY) coated with type I rat tail collagen (150 μg/ml; Sigma). For immunocytochemistry, the cells were grown on collagen type I-coated glass coverslips or collagenous membranes in a transwell configuration (MP Biomedicals, Solon, OH). hCMEC/D3 cells formed confluent, contact-inhibited monolayers, and were used consistently between passages (p)28 and 35.
Growth and development of hCMEC/D3s
To test if SPARC expression changed under varied serum conditions, confluent hCMEC/D3 cultures (p30–32) were grown in complete growth media and replenished with fresh medium containing either no FBS, 1 % FBS, 5 % FBS, or 0.1 % bovine serum albumin (BSA) (Sigma) in place of FBS. hCMEC/D3 growth was assessed for culture confluency where cells were approximately 20–30 % confluent after 1 day in culture (DIC), 50–70 % confluent after 4 DIC, and confluent after 6–7 DIC. Protein and mRNA was extracted from cultures at subconfluent and confluent growth stages and subjected to immunoblotting and semi-quantitative RT-PCR. hCMEC/D3 cultures grown on collagen membranes were fixed in 4 % paraformaldehyde (PFA)-PBS when subconfluent (50–70 % confluent) and confluent, before being probed with monoclonal antibodies against SPARC, ZO-1, and Ki-67 for immunocytochemical analysis.
hCMEC/D3 exposure to SPARC or inflammatory mediators
hCMEC/D3s cultured on collagen type I-coated six-well plates until confluent were treated 1 day later with 0.1, 1, 10 μg/ml rh-SPARC (R&D Systems) and TNF-α (200 U/ml) in fresh reduced serum (1 % FBS) complete media for 24 h. The protein levels of endothelial tight junctions were determined by Western blotting analysis. Confluent monolayers grown on collagen membrane inserts were replenished with complete media containing 10 or 100 U/ml TNF-α, 100, 200, or 500 U/ml IFN-γ, and 50 ng/ml LPS, alone and in combination. For cytokine treatment of cells on collagen membranes, media containing cytokine was added to the upper chamber.
Barrier function assays
Barrier function was examined using TEER impedance measurements and transendothelial FITC-dextran (3 and 10 kDa) diffusion assays. hCMEC/D3 (p30–32) were grown on transwell polyester membrane inserts (0.4-μm porosity and 1.12 cm2 surface area, Corning Costar #3460) in serum-reduced (1 % FBS) complete media until 2 days post-confluence, and the measured TEER had stabilized. Media was replenished into both upper (luminal) and lower (abluminal) chambers; however, cytokine- and SPARC-containing media were only applied into the upper chamber. TEER was measured daily by EVOM-2 ohm meter using ENDOHM-12 chamber and STX2 chopstick electrodes (World-Precision Instruments, Sarasota, FL). Resistance of blank filters without cells was subtracted from those with cells, and these values were multiplied by the surface area of the inserts to get the final resistance (Ω·cm2).
Endothelial monolayer permeability was studied by measuring fluorescently labeled dextran diffusion through a confluent monolayer of hCMEC/D3 cells grown on transwell inserts. These studies were performed by adding 50 μg/ml dextran (3 or 10 kDa) tagged with Alexa Fluor 488 (Molecular probes, Eugene, OR) to the luminal chamber of a transwell insert and sampling (10 μl/sample) from the abluminal chamber at 30, 60, 120, and 180 min into a 96-well plate, an equal volume of fresh medium replaced each time. Dextran fluorescence in the samples was measured by a microplate fluorometer (Floroskan Ascent 374, ThermoScientific, Hudson, NH) at 485 nm (excitation) and 525 nm (emission). Raw values from dextran fluorescence were converted to concentrations using a standard curve and slope of the linear regression line. Permeability coefficient was calculated as described previously [
4].
Cell lysis and Western blotting
To lyse hCMEC/D3 cells grown on six-well plates, the cells were first washed twice with cold PBS, then incubated in situ on ice for 10 min in 125 μl of ice-cold NP-lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, and 1 % NP-40, adjusted to pH 8.0) with fresh protease inhibitor and phosphatase inhibitor cocktails added before use (Roche, Laval, QC). The cells were then scraped and collected into tubes, triturated on ice through a 28-Gauge needle five times and centrifuged at 13,500 rpm for 12 min at 4 °C. The supernatant was collected as whole cell lysate and stored at −80 °C. hCMEC/D3 cell lysis for TJ protein analysis required stronger detergents, 125 μl of ice-cold radio-immunoprecipitate assay (RIPA) buffer containing 20 mM Tris-HCl, 150 mM NaCl, 2 mM EDTA, 1 % NP-40, 0.1 % SDS, and 0.5 % sodium deoxycholate.
Protein concentrations were determined by bicinchoninic acid (BCA) protein assay kit (Sigma). Protein estimations based on the BSA standard absorbance curve. Lysates (20–40 μg) were diluted 1:1 in reducing 2× Laemmli sample buffer (Bio-Rad, CA) containing 0.05 % β-mercaptoethanol and heated at 100 °C for 5 min. The samples were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE: 6 % gel for ZO-1, 12 % gel for SPARC, 15 % gel for occludin and claudin-5) for 1.5 h at 80–100 V and transferred to nitrocellulose membranes in wet transfer buffer (25 mM Tris, 192 mM glycine, 20 % methanol) at 90 mA overnight at 4 °C. Immunoblots for SPARC, claudin-5, occludin, and GAPDH were blocked in 5 % skim milk in Tris-buffered saline (TBS: 1.37 M NaCl, 27 mM KCl, 0.25 M Tris, adjusted to pH 7.4) for 1 h then incubated in primary antibody diluted in 2 % skim milk-TBST (TBS + 0.01 % Tween 20 (TW20)) overnight at 4 °C (or 2 h at RT for GAPDH). After washing, HRP-conjugated secondary antibodies diluted in 2 % skim milk-TBST (TBS + 0.01 % TW20) were added and then again washed in TBST. Immunoblots for detection of ZO-1 were blocked in 1 % BSA-TBST (TBS + 0.05 % TW20) for 4 h at RT and then incubated in monoclonal anti-ZO-1 primary antibody diluted in TBS overnight at 4 °C followed by washing in TBST and incubation with secondary antibody. All immunoblots were developed in enhanced chemiluminescence (ECL) substrate (ThermoFisher Scientific, Bothell, WA). Band signals were detected using a Versadoc imaging system (Bio-Rad Laboratories), and band densities were quantified by ImageJ 1.42i software (National Institutes of Health, Bethesda, MD).
Semi-quantitative RT-PCR
Total RNA was extracted from cell cultures using RNeasy Mini Kit (QIAGEN Science, Toronto, ON) according to the manufacturer’s protocol. Complementary DNA (cDNA) was synthesized by High Capacity RNA-to-cDNA Master Mix Kit (Applied Biosystems, Foster City, CA). Relative quantification of PCR reactions were performed with Platinum PCR SuperMix (Invitrogen) using the following primer sequences for SPARC transcript amplification 5′-AGTGCACCCTGGAGGGCACC-3′ (Forward); 5′-TGCTTGATGCCGAAGCAGCC-3′ (Reverse) and the following primers for the GAPDH housekeeping gene 5′-AAGGCTGGGGCTCATTTGCAG-3′ (Forward); 5′-CTGCTTCACCACCTTCTTGATG-3′ (Reverse). A semi-quantitative analysis of mRNA levels was carried out using scans with the Bio-Rad Gel Doc UV system (Bio-Rad Laboratories, Hercules, CA) and differences in SPARC and GAPDH expression were calculated by Image Lab software (Bio-Rad Laboratories).
Immunocytochemistry
Following co-incubations, cultures were washed twice with warm (37 °C) PBS and fixed at RT in 4 % PFA-PBS for 10 min. Cultures were directly stored in PBS-0.01 % sodium azide at 4 °C until stained. For staining, cultures were washed twice with PBS then incubated for 10 min in permeabilization-blocking buffer (0.1 % Triton X-100 and 4 % normal goat serum (NGS) in PBS). Cultures were incubated twice for 10 min in blocking buffer (4 % NGS-PBS) then incubated with primary antibody diluted in blocking buffer for 1 h at RT or overnight at 4 °C. After primary antibody incubation, cultures were washed three times for 5 min with PBS then incubated in specific secondary antibody in blocking buffer for 50 min in the dark. Cultures were washed twice in PBS, stained with DAPI nuclear stain (1:10,000) in PBS for 5 min, and again washed three times in PBS. Membranes were excised from discs using a scalpel, drained of excess PBS, and embedded in 10 μl of ProLong Gold anti-fade reagent (Invitrogen) underneath a glass coverslip. Cultures on the glass coverslips were mounted cell side down on 10 μl of ProLong Gold.
Image acquisition and analysis
Distribution analysis was performed in a blinded fashion where cells were distinguished and coded numerically on a DAPI image to avoid duplication and evaluated for level of SPARC expression. SPARC immunoreactivity levels were assigned to individual cells according to the classification scheme and representative images in Table
1. Fluorescence images were captured with an Axioplan 2 imaging epifluorescent microscope (Zeiss, Jena, Germany) and Axiovision 4 software (Zeiss). Confocal micrographs were captured with an Olympus Fluoview 1000 laser scanning confocal microscope with Nomarski optics and FV1000 Fluoview software (Olympus Corporation, Tokyo Japan). Analysis was performed using Adobe Photoshop extended CS3 version 10.0 or ImageJ 1.42i software (NIH) or assessed by SPARC immunoreactivity scale. Adobe Photoshop measurements of nuclear Ki-67 intensity were performed by outlining individual nuclei on a DAPI/blue filter image using the “quick selection tool”, and measuring Ki-67/green filter mean pixel intensity (MPI)—the average intensity of all pixels within the DAPI-stained/delineated nuclear regions. Ki-67 positivity was relative to a threshold determined by blindly screening three images of confluent and subconfluent images each and denoting those cells negative for Ki-67. A threshold limit was set by averaging the Ki-67 MPI of those nuclei visually deemed Ki-67 negative (
n = 59) plus three standard deviations. Ki-67-positive nuclei were those exceeding this pre-defined Ki-67 threshold. Quantification for SPARC immunofluorescence was performed either by regional analysis, where regions for analysis were demarcated by thresholding the cell edge such that only the cell-covered surface area of a field/image was assessed for MPI using ImageJ or assessed based on the SPARC immunoreactivity scale. Cell counting of DAPI-positive nuclei was performed using the ImageJ “analyze particles” tool and automated using ImageJ macroscript.
Table 1
SPARC immunoreactivity assessment scale for hCMEC/D3 cells in culture
Statistical analysis
GraphPad Prism version 5.01 (GraphPad Software, San Diego, CA) was used for graph synthesis and statistical analysis. Parametric data expressed as mean ± standard error of the mean (mean ± SEM) were analyzed by one-way analysis of variance (ANOVA) and Newman-Keul multiple comparison test. Non-parametric data expressed as mean ± standard deviation (mean ± SD) were analyzed by a Kruskall-Wallis and Dunn’s multiple comparison test. Mann-Whitney comparison was performed on data comparing two non-parametric groups/treatments. SPARC immunoblots and functional assays were analyzed by one-way ANOVA followed by a Tukey’s honestly significant difference (HSD) multiple comparisons analysis.