Address correspondence and reprint requests to Dr Aloïse Mabondzo, Commissariat à l’Énergie Atomique (CEA), Service de Pharmacologie et d'Immunologie, DRM/DSV, Bâtiment 136, 91191 Gif sur Yvette Cedex, France. E-mail: Aloïse.Mabondzo@cea.fr
Evidence suggests that endothelin-1 (ET-1) plays an essential role in brain inflammation. However, whether ET-1 contributes directly to blood–brain barrier (BBB) breakdown remains to be elucidated. Using an in vitro BBB model consisting of co-cultures of human primary astrocytes and brain microvascular endothelial cells (BMVECs), we first investigated the expression of ET-1 by BMVECs upon stimulation with tumour necrosis factor (TNF)-α, which plays an essential role in the induction and synthesis of ET-1 during systemic inflammatory responses. Increased ET-1 mRNA was detected in the human BMVECs 24 h after TNF-α treatment. This was correlated with an increase in ET-1 levels in the culture medium, as determined by sandwich immunoassay. Both TNF-α and ET-1 increased the permeability of human BMVECs to a paracellular tracer, sucrose, but only in the presence of astrocytes. The increase in BMVEC permeability by TNF-α was partially prevented by antibody neutralization of ET-1 and completely by monoclonal antibody against IL-1β. Concomitantly, TNF-α induced IL-1β mRNA expression by astrocytes in co-culture and this effect was partially prevented by ET-1 antibody neutralization. In parallel experiments, treatment of human primary astrocytes in single cultures with ET-1 for 24 h induced IL-1β mRNA synthesis and IL-1β protein secretion in the cell culture supernatant. Taken together, these results provide evidence for paracrine actions involving ET-1, TNF-α and IL-1β between human astrocytes and BMVECs, which may play a central role in BBB breakdown during CNS inflammation.
Endothelin-1 (ET-1) is a 21-amino-acid peptide considered to belong to the cytokine family. Isolated from endothelial cells, ET-1 has been found to be one of the most potent vasoconstrictor peptides in humans (Yanagisawa et al. 1988). In addition, ET-1 has been implicated as a mediator of cerebrovascular responses in ischaemic stroke and subarachnoid haemorrhage (Masaoka et al. 1989; Kohno et al. 1990; Suzuki et al. 1990; Yasuda et al. 1990; Lampl et al. 1997). ET-1 also exerts a wide spectrum of effects on non-vascular tissues, including those of the CNS. Recent evidence demonstrated the neuropathological significance of ET-1, as shown by the correlation between increased blood–brain barrier (BBB) permeability (Narushima et al. 1999; Chi et al. 2001), and increased ET-1 levels in CSF and CNS parenchyma (Rolinski et al. 1999), although its prime role in this process and the underlying mechanisms remain to be elucidated.
The role of cytokines such as interleukin (IL)-1β and tumour necrosis factor (TNF)-α in the neuropathogenesis of brain inflammation is well established (Griffin et al. 1994; Boven et al. 1999). IL-1β has been shown to be associated with the pathophysiology of demyelinating disorders such as multiple sclerosis and viral infections of the CNS (Clerici et al. 2001; Luomala et al. 2001; Niino et al. 2001). Recent evidence has further demonstrated the pathological significance of IL-1β (Samad et al. 2001). Thus, inflammation is a key component in traumatic brain injury pathophysiology.
As the release of ET-1 can be modulated by inflammatory cytokines such as TNF-α (Ohta et al. 1990; Yoshizumi et al. 1990; Ehrenreich et al. 1993; Kugaya et al. 1997), we first investigated the expression of ET-1 by the major cellular component of the BBB, brain microvascular endothelial cells (BMVECs), upon stimulation with TNF-α. In addition, ET-1-induced effects on BBB permeability were examined. Particular emphasis was placed on the paracrine effects of ET-1 on IL-1β production by human primary astrocytes. Here we show that TNF-α induced the production of ET-1 in an in vitro model of human BBB. Both TNF-α and ET-1 induced permeability changes in human BMVECs cultured above astrocytes but not in a single monolayer culture of BMVECs. This effect was associated with TNF-α-induced IL-1β synthesis by astrocytes which was prevented by ET-1 antibody neutralization. ET-1 treatment of human primary astrocytes in single culture induced IL-1β synthesis in these cells. We therefore suggest that paracrine actions involving ET-1, TNF-α and IL-1β between astrocytes and brain endothelial cells may have a role in BBB breakdown during CNS inflammation.
Materials and methods
The following reagents were purchased: ET-1 from Biovalley (Marne la Vallée, France); ET-1, β-actin, cyclophilin and IL-1β primers from MWG Biotech (Courtabeuf, France); [U-14C]sucrose (350 mCi/mmol) from Amersham Pharmacia Biotech (Amersham, UK); enhanced avian RT–PCR kit and recombinant human TNF-α from Sigma (St Louis, MO, USA); isotypic mouse IgG1 monoclonal antibody from Boehringer Mannheim Biochemica (Meylan, France); R-phycoerythrin-conjugated secondary antibodies from Jackson Immunoresearch (West Baltimore, MD, USA); platinum quantitative PCR SuperMix-UDG from Life Technologies (Noisy le Grand, France); and iCycler PCR Reaction Mix SYBR Green I from Bio-Rad Laboratories (Hercules, CA, USA). Mouse monoclonal antibody (mAb) anti-ET-1 was produced in the laboratory.
In vitro human BBB model
The in vitro BBB model consisted of human BMVECs co-cultured with human primary astrocytes (Bio Whittaker, Emrainville, France) as described previously (Mégard et al. 2002). All cell strains tested negative for human immunodeficiency virus 1 and hepatitis B DNA by PCR. BMVECs and astrocytes stained positively for 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate acetylated low-density lipoprotein and glial fibrillary acid protein respectively. Briefly, astrocytes (5 × 104 cells) were plated on transwell plates (Costar; Dutscher, Brumath, France) in an astrocyte-specific basal medium (Bio whittaker). After 72 h, BMVECs (5 × 104 cells) were seeded on the upper side of a collagen-coated polytetrafluoroethylene polyester polystyrene TranswellR membrane (Costar; pore size 0.4 µm, diameter 12 mm, insert growth area 1 cm2) in the BMVEC-specific medium (0.5 mL). The microplates were then incubated at 37°C in an atmosphere containing 5% CO2. Under these experimental conditions, BMVECs formed a confluent monolayer within 15–20 days (Mégard et al. 2002).
Flow cytometric analysis of cells was performed on a FACS-Calibur cytometer (Becton-Dickinson, Rungis, France). For cytoplasmic staining, cells were scraped and fixed with Cytofix/cytopermeTM solution (Pharmingen, Rungis, France) for 20 min at 4°C. The cells were washed and resuspended in 50 µL perm/wash (Pharmingen) and labelled for 30 min on ice with primary specific antibodies (1 µg/mL) followed by a further 30 min with phycoerythrin-conjugated goat anti-mouse IgG-F(ab′)2 antibody in the presence of human serum (10%). Unstained cells were used as background control. Negative control analyses were carried out using isotype-matched irrelevant antibodies in place of primary antibody.
Exposure of the in vitro human BBB model to TNF-α
On the day of the experiments, inserts containing the monolayer of human BMVECs co-cultured with astrocytes were transferred on to a new plate. Cell monolayers were treated with increasing doses of TNF-α (0.005–0.5 µg/mL) for 3 and/or 24 h. After exposure to TNF-α, the supernatant above the model was removed and frozen for enzymatic immunometric assay of ET-1. The BMVECs comprising the monolayer were washed with phosphate-buffered saline (PBS) and then lysed for total RNA extraction.
Enzyme immunometric assay for ET-1
Immunometric assays using acetylcholinesterase conjugate were performed in 96-well microtitre plates coated with anti-ET-1 monoclonal antibody (mAb Endo-18) as described previously (Créminon et al. 1993). Briefly, 100 µL ET-1 solution (standard or sample) were added to each well with 100 µL enzyme tracer (anti-ET-1 mAb Endo-4) labelled with acetylcholinesterase enzyme used at a concentration of 5 or 10 Ellman units/mL). Immunoreaction was allowed to proceed for various times at 20°C. The plates were then washed and solid phase-bound acetylcholinesterase activity was determined by addition of 200 µL Ellman's reagent and measured at a wavelength of 414 nm.
Enzyme-linked immunosorbent assay (ELISA) for IL-1β in cell culture supernatants
ELISAs for human IL-1β in the astrocyte cell culture supernatants were performed using the OptEIATM human IL-1β set (Pharmingen). Briefly, after added 100 µL of diluted capture antibody to each well, 100 µL standard or sample was added with 100 µL biotinylated anti-human IL-1β polyclonal antibody. Immunoreaction was allowed to proceed for 30 min at room temperature. The plates were then washed and 100 µL avidin–horseradish peroxidase conjugate was added to each well and incubated for 30 min at room temperature. After addition of 100 µL tetramethylbenzidine and hydrogen peroxide substrate solution, the absorbance was read at 450 nm.
RNA extraction and real-time RT–PCR
Total RNA was isolated from the cell samples with GenElute (Mammalian total RNA kit; Sigma) and quantified by optical density. Of the total RNA, 1.25 µg was used for RT with enhanced avian reverse transcriptase (Sigma) according to the manufacturer's instructions. Of the cDNA product, 5 µL was added to 45 µL containing platinum quantitative PCR super Mix UDG [60 U/mL, Platinum Taq DNA polymerase, 40 mm Tris-HCl (pH 8.4), 100 mm KCl, 6 mm MgCl2, 400 µm dGTP, 400 µm dATP, 400 µm dCTP, 800 µm dUTP, 40 U/mL uracil DNA glycosylase (UDG)], each primer at 300 nm and iCycler PCR Reaction Mix SYBR Green I. The primers used and PCR conditions are summarized in Table 1. After amplification, a melting curve was obtained by heating at 55°C and fluorescence data were collected at 0.5°C/s.
Table 1. Nucleotide sequences of probes used in RT–PCR assays and amplification conditions
Sense 5′ GAGAAACCCACTCCCAGTCC 3′ Antisense 5′ GATGTCCAGGTGGCAGAAGT 3′
94°C for 105 s, 60°C for 30 s, 72°C for 30 s; 50 cycles
Sense 5′ ACGGGGTCACCCACACTGTGC 3′ Antisense 5′ CTAGAAGCATTGCGGTGGACGATG 3′
94°C for 105 s, 60°C for 30 s, 72°C for 30 s; 50 cycles
Sense 5′ GGGCCTCAAGGAAAAGAATC 3′ Antisense 5′ CTGCTTGAGAGGTGCTGATG 3′
94°C for 30 s, 60°C for 30 s, 72°C for 45 s; 50 cycles
Quantification of real-time PCR for ET-1
Quantitative analysis of the iCycler iQTM data was performed employing iCycler analysis software (Bio-Rad, Marnes la Coquette, France). The data analysis is divided into two phases: a specificity control for the amplification reaction using the melting curve program of the iCycler iQTM software, followed by use of the quantification program. The SYBR Green I signal of each sample is plotted against the number of cycles. The iCycler iQTM analysis software is used to remove background by setting a noise band. This fluorescence threshold is used to determine cycle numbers that correlate inversely with the log of the initial template concentration. As described by others (Hofmann et al. 2001; Smith et al. 2001), we used the expression of β-actin to normalize the expression data for the ET-1 gene. β-Actin was used as an active and endogenous reference to correct for differences in the amount of total RNA added to a reaction and to compensate for different levels of inhibition during RT of RNA and during PCR.
Endothelial cell permeability
Permeability experiments were carried out as described previously (Mégard et al. 2002). Briefly, after exposure of the in vitro human BBB model to TNF-α or ET-1, on the day of the experiments inserts containing the monolayer of human BMVECs co-cultured with astrocytes were transferred on to a new plate for permeability experiments. Some 0.5 mL BMVEC-specific medium and 1.5 ml astrocyte-specific medium was added to the upper and basal chamber respectively, then 1 µCi/mL (2 nm) [14C]labelled sucrose was added to the upper chamber. The amount of tracer that passed through the inserts at different time points was determined by scintillation counting, and cleared volume with respect to time was calculated as described elsewhere (Pardridge et al. 1990). Triplicate co-cultures and triplicate control wells were used.
Statistical comparisons were made with one-way anova and two-tailed Student's t-test. p < 0.05 was considered statistically significant.
TNF-α induced production of ET-1 in an in vitro co-culture model of the human BBB
To establish ET-1 production by human brain endothelial cells in response to TNF-α treatment, amplification of ET-1 DNA was assessed. To quantify the relative abundance of the ET-1 transcripts, we performed real-time PCR using specific primers (Table 1). Real-time PCR showed that the ET-1 mRNA was 2.7- to 8.5-fold higher in BMVECs treated with TNF-α at 0.1 and 0.5 µg/mL respectively (Figs 1a, c and e), whereas the expression of the control β-actin gene did not differ between untreated and TNF-α-treated brain endothelial cells (Figs 1b and d). To test ET-1 release by endothelial cultures, we used an enzyme immunometric assay to measure the levels in supernatants from untreated and TNF-α-treated brain endothelial cells cultured above astrocytes. We found that the levels of ET-1 were significantly increased by 24 h after treatment (p < 0.001) (Fig. 2).
ET-1 failed to increase brain endothelial permeability in the absence of astrocytes
Inserts containing the monolayer of the human BMVECs co-cultured with astrocytes were transferred on to a new plate. Endothelial cell monolayers in single cultures were then treated with ET-1 (0.1 and 1 µg/mL) and 0.1–0.5 µg/mL TNF-α for 24 h. The permeability to the paracellular marker sucrose across the cell monolayer was then assessed. We found that the clearance value for sucrose was not significantly different between untreated and treated cultures (Figs 3a and b). However, treatment of the monolayer of human BMVECs with ET-1 or TNF-α at 0.1 and 0.5 µg/mL for 24 h in the presence of astrocytes in the basal chamber significantly increased the permeability of the human BMVECs (Figs 4a and b). With antibody neutralization of ET-1 (10 µg/mL), the increase in BMVEC permeability mediated by TNF-α at 0.5 µg/mL was partially prevented (Fig. 4b).
Exposure of the in vitro human BBB model to TNF-α induced IL-1β mRNA synthesis by astrocytes
To demonstrate the possible participation of ET-1 in BBB disruption, we investigated whether ET-1 produced by the BMVECs in response to TNF-α caused BBB disruption by mechanisms involving IL-1β production. The monolayer of human BMVECs was treated with 0.5 µg/mL TNF-α in the upper chamber. Twenty-four hours after cell treatment, astrocytes in the basal chamber were removed, washed with PBS and then lysed for total RNA extraction. Amplification of IL-1β cDNA was assessed by real-time RT–PCR using the specific primers (Table 1). We found that predominant IL-1β mRNA was 12.5-fold higher in astrocytes when co-cultures were treated with TNF-α. Pretreatment of astrocytes with ET-1 antibody neutralization (10 µg/mL) for 1 h partially prevented IL-1β mRNA synthesis in these cells (Fig. 5).
ET-1 increased expression of IL-1β in single cultures of human primary astrocytes
Human primary astrocytes were treated for 24 h with ET-1 (0.5 µg/mL). Real-time RT–PCR showed that ET-1 induced the overexpression of IL-1β mRNA (fivefold higher) in human primary astrocytes after 24 h of treatment (Figs 6a, b and d), whereas the expression of the control β-actin gene did not differ between untreated and ET-1-treated astrocytes (Fig. 6c). To test for IL-1β secretion by astrocyte cultures, we used an ELISA to measure the levels in supernatants from untreated and ET-1-treated astrocytes. We found that the levels of IL-1β had significantly increased by 24 h after treatment with 0.5 µg/mL ET-1 (p < 0.0001) (Fig. 7).
Effect of IL-1β on BBB permeability
The effects of IL-1β on BBB permeability were investigated. IL-1β (1 ng/mL) was added to the basal chamber of the experimental system for 24 h. The permeability of the paracellular marker sucrose across human BMVECs co-cultured with astrocytes was then analysed. We found that the cleared volume for sucrose was significantly increased in treated compared with untreated cells (p < 0.0001) (Fig. 8). In order to implicate IL-1β released from astrocytes in the increase in BBB permeability induced by ET-1, blockade experiments were performed with IL-1β monoclonal antibody. Astrocytes were pretreated with 10 µg/mL IL-1β monoclonal antibody for 1 h in the basal chamber before treatment with ET-1 (0.5 µg/mL) for 24 h. The permeability to sucrose across the cell monolayer of human BMVECs was then assessed as described above. Pretreatment of astrocytes with IL-1β antibody for 1 h prevented the increased in BBB permeability induced by ET-1 (p < 0.0001) (Fig. 8).
The purpose of this study was to evaluate the ability of ET-1 to mediate in vitro BBB breakdown during CNS inflammation and to investigate the underlying mechanisms. We investigated the hypothesis that ET-1 produced by the BMVECS in response to systemic inflammatory responses leads to BBB disruption by mechanisms involving IL-1β synthesis by astrocytes. This was carried out in an in vitro model of the human BBB (Mégard et al. 2002) that exhibits important features of the BBB in vivo. Major findings of this study are that: (i) the pro-inflammatory cytokine TNF-α significantly increased ET-1 mRNA expression by human BMVECS, an effect correlated with an increase in ET-1 levels in the culture supernatant; (ii) both TNF-α and ET-1 increased the permeability of human BMVECS to a paracellular marker (sucrose), but only in the presence of human primary astrocytes; (iii) the increase in BMVEC permeability was partially prevented by antibody neutralization of ET-1; (iv) this effect was associated with TNF-α-induced IL-1β synthesis by human primary astrocytes; (v) TNF-α-induced IL-1β synthesis in astrocytes was mediated by ET-1; (vi) ET-1 induced IL-1β synthesis in these cells; and (vii) IL-1β increased the permeability of BMVECS. Taken together, these findings provide evidence for an immunoregulatory loop between ET-1, TNF-α and IL-1β in the observed alterations in BBB function during CNS inflammation. These findings are discussed below.
It has been shown that the pro-inflammatory cytokine TNF-α contributes to the breakdown of the BBB in the pathogenesis of a number of brain disorders. For example, TNF-α has been found to alter the permeability of bovine brain endothelial cells (Deli et al. 1995) and cause changes in the BBB in vivo (Ramilo et al. 1990; Kim et al. 1992). The same effect has been described in monolayers of rat cerebral endothelial cells (De Vries et al. 1996). However, the interaction of ET-1 and TNF-α in this process has not been demonstrated using an in vitro primary cell preparation (BMVECs and human primary astrocytes) that mimics as closely as possible the normal human BBB.
TNF-α, which was present in the upper chamber of our experimental system, induced ET-1 production in a dose-dependent manner at the level of the human BBB. Neither TNF-α nor ET-1 directly increased brain endothelial cell permeability. The permeability of treated BMVECs was not statistically different from that of untreated cells. This shows that the effect of TNF-α and ET-1 described previously on BBB permeability was not direct (Burke-Gaffney and Keenan 1993; De Vries et al. 1996; Miller et al. 1996) but rather involved alternative mechanisms. These may include the involvement of parenchymal cellular elements of the brain, as the intracisternal injection of TNF-α in newborn piglets (Megyeri et al. 1992) and the intracisternal application of ET-1 in the dog mediate disruption of the BBB (Narushima et al. 1999). The presence of astrocytes under the monolayer of brain endothelial cells shows the importance of the immunoregulatory loop in the BBB disruption process. Indeed, treatment of BMVECs with TNF-α and/or ET-1 in the presence of human primary astrocytes significantly increased their permeability. Both concentrations of TNF-α (0.1 and 0.5 µg/mL) induced increases in BMVEC permeability with the maximal effect at the highest dose tested (Fig. 4b). The brain endothelial cell permeability induced by TNF-α was partially abrogated by antibody neutralization of ET-1. These findings suggest that ET-1 functions as a mediator between brain endothelial cells and astrocytes, and participates in the BBB breakdown process. We showed a significant dose-dependent increase in BMVEC permeability with TNF-α. Our results suggest mechanisms different from those proposed for BBB breakdown during endothelial cell monolayer treatment with 0.1 µg/mL TNF-α (Brett et al. 1989; Karen et al. 2001; Mark and Muller 1999; Mark et al. 2001), such as G protein-mediated conformational changes in the actin-based cytoskeleton (Brett et al. 1989) and prostaglandin involvement (Karen et al. 2001).
Exposure of BMVECs to TNF-α induced an approximately 12.5-fold increase in IL-1β mRNA synthesis by astrocytes. This effect was partially prevented by ET-1 antibody neutralization. This finding suggests the paracrine action of ET-1 in this process in response to TNF-α by brain endothelial cells. These results agree with the finding of an induction of IL-1β mRNA synthesis in a single culture of human primary astrocytes treated with ET-1 (fivefold above that in untreated astrocytes; Fig. 6d). This is the first report finding of ET-1-induced IL-1β over-expression in human primary astrocytes. Moreover, blockade experiments with the neutralizing monoclonal antibody against IL-1β (Fig. 8) demonstrated that IL-1β is a prime determinant in the increase in brain endothelial cell permeability. This is consistent with previous work that demonstrated a major role of IL-1β in inducing changes in BBB permeability in the rat (Blamire et al. 2000).
In conclusion, our in vitro studies have demonstrated paracrine actions involving ET-1, TNF-α and IL-1β between human primary astrocytes and BMVECs which may play a pivotal role in breakdown of the BBB.
We would like to thank the Agence National de Recherche sur le SIDA for financial support.