A pericyte-derived angiopoietin-1 multimeric complex induces occludin gene expression in brain capillary endothelial cells through Tie-2 activation in vitro

Authors

  • Satoko Hori,

    1. Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Aoba-ku, Sendai, Japan
    2. New Industry Creation Hatchery Center, Tohoku University, Aoba-ku, Sendai, Japan
    3. CREST and SORST of the Japan Science and Technology Agency (JST), Japan
    Search for more papers by this author
  • Sumio Ohtsuki,

    1. Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Aoba-ku, Sendai, Japan
    2. New Industry Creation Hatchery Center, Tohoku University, Aoba-ku, Sendai, Japan
    3. CREST and SORST of the Japan Science and Technology Agency (JST), Japan
    Search for more papers by this author
  • Ken-ichi Hosoya,

    1. Faculty of Pharmaceutical Sciences, Toyama Medical and Pharmaceutical University, Toyama, Japan
    2. CREST and SORST of the Japan Science and Technology Agency (JST), Japan
    Search for more papers by this author
  • Emi Nakashima,

    1. Department of Pharmaceutics, Kyoritsu College of Pharmacy, Minato-ku, Tokyo, Japan
    2. CREST and SORST of the Japan Science and Technology Agency (JST), Japan
    Search for more papers by this author
  • Tetsuya Terasaki

    1. Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Aoba-ku, Sendai, Japan
    2. New Industry Creation Hatchery Center, Tohoku University, Aoba-ku, Sendai, Japan
    3. CREST and SORST of the Japan Science and Technology Agency (JST), Japan
    Search for more papers by this author

Address correspondence and reprint requests to Professor Tetsuya Terasaki, Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Aoba, Aramaki, Aoba-ku, Sendai 980–8578, Japan.
E-mail: terasaki@mail.pharm.tohoku.ac.jp

Abstract

Although tight-junctions (TJs) at the blood–brain barrier (BBB) are important to prevent non-specific entry of compounds into the CNS, molecular mechanisms regulating TJ maintenance remain still unclear. The purpose of this study was therefore to identify molecules, which regulate occludin expression, derived from astrocytes and pericytes that ensheathe brain microvessels by using conditionally immortalized adult rat brain capillary endothelial (TR-BBB13), type II astrocyte (TR-AST4) and brain pericyte (TR-PCT1) cell lines. Transfilter co-culture with TR-AST4 cells, and exposure to conditioned medium of TR-AST4 cells (AST-CM) or TR-PCT1 cells (PCT-CM) increased occludin mRNA in TR-BBB13 cells. PCT-CM-induced occludin up-regulation was significantly inhibited by an angiopoietin-1-neutralizing antibody, whereas the up-regulation by AST-CM was not. Immunoprecipitation and western blot analyses confirmed that multimeric angiopoietin-1 is secreted from TR-PCT1 cells, and induces occludin mRNA, acting through tyrosine phosphorylation of Tie-2 in TR-BBB13 cells. A fractionated AST-CM study revealed that factors in the molecular weight range of 30–100 kDa led to occludin induction. Conversely, occludin mRNA was reduced by transforming growth factor β1, the mRNA of which was up-regulated in TR-AST4 cells following hypoxic treatment. In conclusion, in vitro BBB model studies revealed that the pericyte-derived multimeric angiopoietin-1/Tie-2 pathway induces occludin expression.

Abbreviations used
AST (PCT)-CM

conditioned medium of TR-AST (TR-PCT) cells

BBB

blood–brain barrier

BCEC

brain capillary endothelial cell

CM

conditioned medium

ECGF

endothelial cell growth factor

JAM

junctional adhesion molecule

PI3-kinase

phosphatidylinositol 3-kinase

PTyr

phosphotyrosine

TEER

transendothelial electrical resistance

TGF-β1

transforming growth factor β1

TJ

tight-junction

TR-AST

conditionally immortalized rat astrocyte cell line

TR-BBB

conditionally immortalized rat brain capillary endothelial cell line

TR-PCT

conditionally immortalized rat pericyte cell line

VEGF

vascular endothelial growth factor

Tight-junctions (TJs) form barriers between adjacent brain capillary endothelial cells (BCECs) at the blood–brain barrier (BBB) and play an important role in preventing non-specific paracellular transport in order to protect the CNS. Brain disorders, such as brain tumors, infarcts and encephalitis, cause TJ disruption to allow BBB leakage (Davies 2002). Therefore, clarifying the mechanism of TJ maintenance is important for understanding and treating CNS diseases associated with BBB leakage.

BCECs are surrounded by pericytes and astrocyte foot processes. The overall brain microvascular biology is a function of the paracrine interactions between BCECs and the two other types of cells (Pardridge 1999; Gaillard et al. 2000; Abbott 2002). Astrocytes are known to induce TJs of non-CNS endothelial cells in vivo (Janzer and Raff 1987). As for brain pericytes, their recruitment and association with microvessels are also key processes in normal vascular development and maintenance. It has been shown that platelet-derived growth factor-B knock-out mice lack brain pericytes, and die due to hemorrhage (Lindahl et al. 1997). Therefore, it is conceivable that paracrine interactions between BCECs, and astrocytes and pericytes, play important roles in maintaining TJs at the BBB.

Occludin (Furuse et al. 1993; Hirase et al. 1997), which is a transmembrane protein, is exclusively localized at the TJ strands of the BBB, together with junctional adhesion molecule (JAM) (Martin-Padura et al. 1998), claudin-5 and claudin-12 (Nitta et al. 2003). These transmembrane proteins are important in regulating the paracellular permeability by interacting with each other in both a homophilic and heterophilic manner between adjacent cells, and by becoming linked to the cytoskelton through a complex of accessory proteins such as ZO-1/ZO-2/ZO-3. Since occludin expression is down-regulated in various brain disorders accompanied by TJ disruption (Huber et al. 2001; Davies 2002), the expression level of occludin is important for TJ maintenance at the mature BBB. Therefore, identifying physiological occludin-regulators will provide deeper insight into the maintenance and recovery of the TJ properties at the BBB.

Angiopoietin-1, a ligand of tyrosine kinase Tie-2, is known to be an anti-permeability factor in the peripheral vascular system (Davis et al. 1996). It has recently been reported that administration of angiopoietin-1 reduces BBB leakage in the ischemic brain (Zhang et al. 2002), suggesting that angiopoietin-1 has an anti-permeability effect on the BBB as well as the peripheral vascular system. In contrast, transforming growth factor β1 (TGF-β1) and vascular endothelial growth factor (VEGF) are vascular permeability factors, and have been reported to be increased in the brain in various neurodegenerative diseases (Kalaria et al. 1998; Lesne et al. 2002). We hypothesize that these soluble factors are secreted from astrocytes and/or brain pericytes in the mature CNS, and involved in TJ regulation at the BBB under physiological and pathophysiological conditions.

Induction and maintenance of the TJ properties at the BBB depend critically on the local conditions and maturational state. In the present study, in order to clarify the astrocyte- and pericyte-derived factors involved in TJ maintenance, we selected conditionally immortalized brain endothelial (TR-BBB) (Hosoya et al. 2000b), type II astrocyte (TR-AST) (Tetsuka et al. 2001), and brain pericyte (TR-PCT) (Asashima et al. 2002; Asashima et al. 2003) cell lines, which had been established from adult transgenic rats harboring temperature-sensitive simian virus 40 large T-antigen (Obinata 1997; Takahashi et al. 1999) and which retain their in vivo function well (Terasaki and Hosoya 2001; Terasaki et al. 2003).

The purpose of the present study was to clarify the mechanism of occludin induction and identify occludin-inducing molecules by using conditionally immortalized BBB cell lines, which are of the same maturational stage, strain, and genetic background. The change in occludin expression level was quantified in TR-BBB13 cells during transfilter co-culture with TR-AST4 cells and on treatment with the conditioned medium of TR-AST4 cells (AST-CM) or TR-PCT1 cells (PCT-CM). We also investigated the effect of angiopoietin-1 in conditioned medium, TGF-β1 and VEGF on occludin expression, and Tie-2 activation by PCT-CM to clarify the signal pathway for occludin regulation.

Materials and methods

Animals

Male Wistar rats, weighing 250–300 g, were purchased from Charles River (Yokohama, Japan). The investigations using rats described in this report conformed to the guidelines established by the Animal Care Committee, Graduate School of Pharmaceutical Sciences, Tohoku University.

Reagents

Endothelial cell growth factor (ECGF) was purchased from Boehringer Mannheim (Mannheim, Germany); benzylpenicillin potassium and streptomycin sulfate were purchased from Wako Pure Chemical Industries (Osaka, Japan); recombinant human angiopoietin-1 was purchased from Genzyme Techne (Minneapolis, MN, USA); recombinant human VEGF and TGF-β1 were purchased from Peprotech EC (London, UK). All other chemicals were commercial products of analytical grade.

Cell cultures

TR-BBB13, TR-AST4 and TR-PCT1 cells were conditionally immortalized BCEC, astrocyte and pericyte cell lines, respectively (Hosoya et al. 2000b; Tetsuka et al. 2001; Asashima et al. 2002) and used as in vitro BBB model (Terasaki et al. 2003). TR-BBB13 cells were grown in Dulbecco's modified Eagle's medium (DMEM, Nissui Pharmaceutical, Tokyo, Japan) supplemented with 20 mm sodium bicarbonate, 15 ng/mL ECGF, 100 U/mL benzylpenicillin potassium, 100 µg/mL streptomycin sulfate and 10% fetal bovine serum (Moregate, Bulimba, Australia) (culture medium-A). The culture medium-B for TR-AST4 cells and TR-PCT1 cells consisted of culture medium-A without ECGF. TR-BBB13 cells and TR-PCT1 cells were seeded onto rat tail collagen type I-coated tissue culture dishes (BD Biosciences, Franklin Lakes, NJ, USA). These cells were maintained at 33°C, which is a permissive temperature at which temperature-sensitive SV40 large T-antigen is activated, in a humidified atmosphere of 95% air and 5% CO2. The experimental culture temperature was also 33°C because a long-term culture needs the cell growth conditions. The cells retain the expression of specific markers at 33°C (Hosoya et al. 2000b; Tetsuka et al. 2001; Asashima et al. 2002).

Transfilter co-culture of TR-BBB13 cells and TR-AST4 cells

TR-BBB13 cells were cultured with TR-AST4 cells in a transfilter co-culture system. In this system, TR-AST4 cells were seeded [5 × 104 cells per insert (4.3 cm2)] on the backside membrane of a collagen type I-coated transfilter, a cell culture insert (pore size: 3.0 µm, BD Biosciences) in culture medium-B. After 24-h culture, the insert was transferred to a 6-well plate and TR-BBB13 cells were seeded (5 × 104 cells per insert) on the upper side of the insert. In single culture, TR-BBB13 cells were seeded (5 × 104 cells per insert) on the upper side of the insert without TR-AST4 cells on the backside. The cells were cultured at 33°C and the culture medium-B was renewed every other day. After a pre-determined time period, transendothelial electrical resistance (TEER) in TR-BBB13 cells was measured using Millicell-ERS equipment (Millipore, Bedford, MA, USA), and after that the cells were collected with a cell scraper.

Preparation of AST-CM and PCT-CM

TR-AST4 cells or TR-PCT1 cells were cultured in culture medium-B without serum. After 24 h, conditioned medium (CM) was collected, concentrated up to 20-fold in a Centriprep-10 (10-kDa cut-off) (Millipore, Bedford, MA, USA) and stored at −20°C until studied. Control CM was prepared by the same procedure using culture medium-B without serum. The CM was adjusted to an appropriate concentration by diluting the 20-fold concentrated CM with culture medium-B without serum.

Fractionation of CM using size exclusion membranes

CM was concentrated up to 20-fold using a graded series of molecular weight cut-off filters as follows (Millipore). The CM was concentrated using a Centriplus-100 (100-kDa cut-off). The retentate was kept as the > 100-kDa fraction and the filtrate was sequentially concentrated using a Centriprep-50 (50-kDa cut-off) (to obtain a 50- to 100-kDa fraction), Centriprep-30 (30-kDa cut-off) (30- to 50-kDa fraction) and Centriprep-10 (10- to 30-kDa fraction).

Treatment with AST-CM, PCT-CM, angiopoietin-1, VEGF or TGF-β1

TR-BBB13 cells were treated with AST-CM, PCT-CM or fractionated AST-CMs for 24 h at 33°C. TR-BBB13 cells were treated with angiopoietin-1 (0, 0.1, 1, 10, 100 and 500 ng/mL), VEGF (0, 0.1, 1 and 10 ng/mL) or TGF-β1 (0, 0.1, 1 and 10 ng/mL) for 24 h at 33°C.

Hypoxic conditions

Hypoxic conditions were achieved with an anaerobic chamber and BBL GasPak Plus (BD Biosciences), which catalytically reduces the oxygen level to less than 10 p.p.m. within 90 min (Shimizu et al. 1996). TR-AST4 cells were cultured for 24 h under these hypoxic conditions.

Angiopoietin-1 inhibitory study

AST-CM and PCT-CM were pre-treated with 0.625 µg/mL antibody against angiopoietin-1 (Chemicon, Temecula, CA, USA) or normal rabbit IgG (control) for 16 h at 4°C. TR-BBB13 cells were cultured with the angiopoietin-1-neutralized-AST-CM or PCT-CM for 24 h at 33°C.

RT-PCR analysis

Total RNA was extracted from rat tissues, TR-AST4 cells and TR-PCT1 cells using an RNeasy kit (Qiagen, Tokyo, Japan) according to the manufacturer's protocol. Single-stranded cDNA was made from 1 µg total RNA by RT (ReverTraAce, Toyobo, Osaka, Japan) using oligo dT primer. The sequences of primers were as follows: sense primer 5′-AGGAGACGGAATACAGGGCT-3′ and antisense primer 5′-CCGGGTTGTGTTGGTTGTAG-3′ for TGF-β1 (GenBank Accession Number; NM021578); sense primer 5′-CTCAGTGGCTGCAAAAACTTG-3′ and antisense primer 5′-CAGAATTTCATTTGTCTGTTGGA-3′ for angiopoietin-1 (GenBank Accession Number; AB080023); sense primer 5′-TTTGAGACCTTCAACACCCC-3′ and antisense primer 5′-ATAGCTCTTCTCCAGGGAGG-3′ for β-actin (GenBank Accession Number; NM031144). The PCR was performed using GeneAmp (PCR system 9700, Perkin-Elmer, Norwalk, CT, USA) with TGF-β1, angiopoietin-1 and β-actin specific primers through 30 cycles of denaturation for 30 s at 94°C, annealing for 30 s at 60°C, and synthesis for 1 min at 72°C. The sizes of the expected RT-PCR products of TGF-β1, angiopoietin-1 and β-actin were 416, 275 and 352 bp, respectively. The RT-PCR of each sample RNA without the RT was used as a negative control. The RT-PCR products were separated by electrophoresis on an agarose gel in the presence of ethidium bromide (0.6 µg/mL) and visualized using an imager (EPIPRO 7000; Aisin, Aichi, Japan). The PCR products were subcloned into a plasmid vector using pGEM-T Easy Vector System I (Promega, Madison, WI, USA) and then sequenced from both directions using a DNA sequencer (CEQ2000XL DNA Analysis System; Beckman Coulter, Fullerton, CA, USA). Sequence comparisons were made using the GENETYX software package, version 6.1.0 (Genetyx, Tokyo, Japan).

Quantitative real-time PCR analysis

Total RNA was extracted from TR-BBB13 cells using an RNeasy kit according to the manufacturer's protocol. RNA integrity was checked by electrophoresis on an agarose gel. Quantitative real-time PCR analysis was performed using an ABI PRISM 7700 sequence detector system (PE Applied Biosystems, Foster City, CA, USA) with 2× SYBR Green PCR Master Mix (PE Applied Biosystems) as per the manufacturer's protocol. To quantify the amount of specific mRNA in the samples, a standard curve was generated for each run using pGEM-T Easy Vector containing occludin, JAM or β-actin (dilution ranging from 0.1 fg/µL to 1 ng/µL). This enabled standardization of the initial mRNA content of TR-BBB13 cells relative to the quantity of β-actin. The control lacking the RT enzyme was assayed in parallel to monitor any possible genomic contamination. PCR was performed through 40 cycles of 95°C for 30 s, 60°C for 1 min, and 72°C for 1 min after pre-incubation at 95°C for 10 min using specific primers. The sequences of primers were as follows: sense primer 5′-GCCTTTTGCTTCATCGCTTCC-3′ and antisense primer 5′-AACAATGATTAAAGCAAAAGCCAC-3′ for occludin (GenBank Accession Number, AB016425); sense primer 5′-ACAGCCATGAGGTCAGAGGCT-3′ and antisense primer 5′-ACCTAGAAGACATTGAAGGCATC-3′ for JAM (GenBank Accession Number, AF276998). The β-actin primers are given above.

Western blot analysis

The membrane and whole cell lysate fractions of rat brain, isolated rat brain capillary, and TR-BBB13 cells were prepared using the procedure described in a previous report (Hosoya et al. 2000a). The AST-CM or PCT-CM was suspended in 10% trichloroacetic acid/acetone, with or without 20 mm dithiothreitol for 1 h at −20°C, followed by centrifugation at 15 000 g for 15 min. The pellet was washed with acetone, with or without 20 mm dithiothreitol, then lyzed with lysis buffer (10 mm Tris-HCl pH 7.4, 1% sodium dodecyl sulfate, 1 mm EDTA and 10% glycerol), with or without 5% 2-mercaptoethanol. Protein concentrations were determined by a DC protein assay kit (Bio-Rad, Hercules, CA, USA). Membrane lysate (20 µg per lane) was used for detecting occludin protein in rat tissues and the cell lines (Fig. 1b). Whole cell lysate (50 µg per lane) was used for investigating the effect of the transfilter co-culture on occludin expression (Fig. 1c). Protein samples were electrophoresed on gradient sodium dodecyl sulfate–polyacrylamide gel (Bio-Rad) and subsequently electrotransferred to nitrocellulose membranes. Membranes were treated with blocking buffer (4% skimmed milk in 25 mm Tris-HCl pH 8.0, 125 mm NaCl, 0.1% Tween 20) for 2 h at room temperature and incubated with anti-occludin antibody (0.1 µg/mL; Zymed, San Francisco, CA, USA), anti-β-actin antibody (1 : 2000; Sigma, St.Louis, MO, USA), anti-angiopoietin-1 antibody (1 µg/mL; Santa Cruz Biotechnology, Santa Cruz, CA, USA) or anti-Tie-2 antibody (1 µg/mL; Santa Cruz Biotechnology) for 16 h at 4°C as the primary antibody. The membranes were washed three times with blocking buffer and incubated with horseradish peroxidase-conjugated second antibody. The bands were visualized with an enhanced chemiluminescence kit (SuperSignal; Pierce, Rockford, IL, USA). The relative densities of the bands were measured using NIH image software (National Institutes of Health, Bethesda, MD, USA).

Figure 1.

Induction of occludin expression in TR-BBB13 cells transfilter co-cultured with TR-AST4 cells. (a) TR-BBB13 cells were transfilter co-cultured with TR-AST4 cells (black column) or single-cultured (open column) for 6 or 8 days. The occludin and JAM mRNA levels were determined by quantitative real-time PCR analysis. Each mRNA expression level was normalized with respect to the β-actin mRNA expression. Each column represents the mean ± SEM (n = 3). **p < 0.01, *p < 0.05, significantly different from the single culture. (b) Western blot analysis of occludin in rat brain and isolated rat brain capillary (positive controls), TR-BBB13 cells and TR-AST4 cells (a negative control). (c) Western blot analysis of occludin (upper) and β-actin (lower) in single-cultured TR-BBB13 cells and TR-BBB13 cells co-cultured with TR-AST4 cells for 8 days. The ratio of occludin to β-actin density in co-culture was 1.77-fold greater than that in single culture (n = 3).

Analysis of tyrosine phosphorylation of Tie-2 protein

TR-BBB13 cells were cultured with two-fold concentrated AST-CM, PCT-CM or 300 ng/mL angiopoietin-1 for 24 h, then collected and suspended in lysis buffer (10 mm Tris-HCl pH 7.5, 1% Triton X-100, 0.5% Nonidet P-40, 1 mm EDTA, 150 mm NaCl, 5 mm sodium pyrophosphate, 10 mmp-nitrophenyl phosphate, 10 mmβ-glycerophosphate, 50 mm sodium fluoride and 1 mm sodium orthovanadate) for 30 min on ice, followed by centrifugation at 15 000 g for 15 min. The Tie-2 protein was immunoprecipitated from the cell lysate using protein Glutathione Sepharose 4B gel beads (Amersham Biosciences, Picataway, NJ, USA) coated with anti-Tie-2 antibody. After electrophoresis under reducing conditions using a gradient gel polyacrylamide (Bio-Rad), the immunoprecipitated Tie-2 proteins were transblotted onto a nitrocellulose membrane. The membrane was incubated with either 1 µg/mL anti-Tie-2 antibody or anti-phosphotyrosine (PTyr) antibody in a blocking buffer for 1 h. The membrane was washed and incubated with horseradish peroxidase-conjugated IgG, then developed as described above. The relative densities of the bands were measured using NIH image software (National Institutes of Health).

Data analysis

Unless otherwise indicated, all data represent the mean ± SEM. An unpaired, two-tailed Student's t-test was used to determine the significance of differences between two groups means. One-way anova followed by the modified Fisher's least-squares difference method was used to assess statistical significance of differences among means of more than two groups.

Results

Induction of occludin expression in TR-BBB13 cells by transfilter co-culture with TR-AST4 cells

The effect of transfilter co-culture with TR-AST4 cells on the expression levels of occludin and JAM was examined in TR-BBB13 cells (Fig. 1). The cells were co-cultured for 6 and 8 days, since it has been reported that cell-to-cell contact was observed between endothelial cells and astrocytes using the same transfilter pore size membrane for over 4 days (Hayashi et al. 1997). As shown in Fig. 1(a), the occludin mRNA levels in 6- and 8-day co-cultured TR-BBB13 cells were significantly increased compared with single culture (9.14- and 1.91-fold, respectively). In contrast, the JAM mRNA level was not changed significantly at either time point. To clarify whether occludin protein was increased concomitantly with the induction of occludin mRNA, the expression of occludin protein was examined in TR-BBB13 cells for 8 days (Figs 1b and c). A single band at 65 kDa was detected in TR-BBB13 cells, corresponding in size to the bands from rat brain and isolated rat brain capillary used as positive controls, whereas no band was detected in TR-AST4 cells (Fig. 1b). Moreover, TR-AST4 cells did not express occludin mRNA (data not shown). The ratio of occludin to β-actin density in co-culture was 1.77-fold greater than that in single culture (Fig. 1c). The TEER in TR-BBB13 cells was unchanged by co-culture (106 ± 2 and 106 ± 1 ohm cm2) compared with single culture (103 ± 1 and 109 ± 2 ohm cm2) for 6 and 8 days, respectively.

Induction of occludin mRNA in TR-BBB13 cells by treatment with AST-CM or PCT-CM

To determine the regulatory effects of soluble factors secreted from TR-AST4 and TR-PCT1 cells, the expressional change of occludin and JAM mRNAs was examined in TR-BBB13 cells treated with AST-CM or PCT-CM (Fig. 2). The occludin mRNA level in TR-BBB13 cells was significantly increased by treatment with both CMs (Fig. 2a). Following treatment with AST-CM, the occludin mRNA level was increased in a concentration-dependent manner, and the occludin mRNA level was increased and reached a plateau at two-fold concentrated CM in the case of PCT-CM treatment. In contrast, the JAM mRNA level was not significantly changed by the presence of one- to five-fold concentrated AST-CM or PCT-CM compared with each control medium (Fig. 2b).

Figure 2.

Induction of occludin mRNA in TR-BBB13 cells by treatment with AST-CM or PCT-CM. Effects of AST-CM (•) or PCT-CM (▪) on the occludin (a) and JAM (b) mRNA levels in TR-BBB13 cells for 24 h. ○, control medium for AST-CM; □, control medium for PCT-CM. The occludin and JAM mRNA levels were determined by quantitative real-time PCR analysis. Each point represents the mean ± SEM (n = 3). Each mRNA expression level was normalized with respect to the β-actin mRNA expression. **p < 0.01, significantly different from the control.

Regulation of occludin mRNA by angiopoietin-1, VEGF and TGF-β1

The effects of 24-h treatment of angiopoietin-1, VEGF and TGF-β1 on the occludin mRNA level were examined in TR-BBB13 cells (Fig. 3). The occludin mRNA level significantly increased following treatment with > 1.0 ng/mL angiopoietin-1 compared with non-treated cells (1.73-fold by 1.0 ng/mL; 1.76-fold by 10 ng/mL; 2.09-fold by 100 ng/mL; 2.20-fold by 500 ng/mL) (Fig. 3a). Conversely, the occludin mRNA level was significantly reduced following the addition of VEGF (1.0 ng/mL, by 25.9%; 10 ng/mL, by 35.9%) or TGF-β1 (0.1 ng/mL, by 17.1%; 1.0 ng/mL, by 46.3%; 10 ng/mL, by 57.4%) for 24 h (Fig. 3a). The JAM mRNA level remained unchanged under any of these conditions (data not shown). To clarify the involvement of TGF-β1 under pathophysiological conditions, the effect of hypoxic conditions on the mRNA expression level of TGF-β1 was determined in TR-AST4 cells by RT-PCR analysis. The mRNA expression of TGF-β1 was increased in TR-AST4 cells following a 24-h period of hypoxia as shown in Fig. 3(b).

Figure 3.

Effects of angiopoietin-1, VEGF and TGF-β1 on the occludin mRNA level in TR-BBB13 cells (a) and effect of hypoxia on the expression level of TGF-β1 mRNA in TR-AST4 cells (b). (a) Effects of angiopoietin-1 (•), TGF-β1 (▴) and VEGF (▪) on the occludin mRNA level in TR-BBB13 cells for 24 h. The occludin mRNA levels were determined by quantitative real-time PCR analysis. Each mRNA expression level was normalized with respect to the β-actin mRNA expression. **p < 0.01, significantly different from the control. (b) RT-PCR analysis of TGF-β1 and β-actin in TR-AST4 cells under normal conditions (control) and 24-h hypoxic conditions. *Respective RT(–) for left-hand lane.

Effect of anti-angiopoietin-1 antibody on induction of occludin mRNA in TR-BBB13 cells by AST-CM and PCT-CM

To clarify the contribution of angiopoietin-1 to the induction of occludin mRNA in TR-BBB13 cells by AST-CM and PCT-CM, angiopoietin-1 activity was neutralized with anti-angiopoietin-1 antibody. The induction of occludin mRNA after 24-h treatment with 10 ng/mL angiopoietin-1 was reduced by 79.0% following pre-treatment with the antibody, indicating that the antibody possesses angiopoietin-1-neutralizing activity (Fig. 4a). The occludin mRNA level was not inhibited in TR-BBB13 cells by AST-CM pre-treated with the antibody (Fig. 4b). In contrast, following stimulation with one-, two- and five-fold concentrated PCT-CM pre-treated with anti-angiopoietin-1 antibody, the occludin mRNA level was inhibited by 62.6, 61.4 and 45.1%, respectively (Fig. 4c).

Figure 4.

Effect of anti-angiopoietin-1 antibody on occludin induction by treatment with AST-CM or PCT-CM. (a) TR-BBB13 cells were cultured in the presence [Ang-1(+)] or absence (control) of recombinant human angiopoietin-1 pre-treated with anti-angiopoietin-1 antibody (open columns) or normal rabbit IgG (black columns). (b) TR-BBB13 cells were cultured with AST-CM (circle symbols) or control medium (diamond symbols) pre-treated with anti-angiopoietin-1 antibody (open symbols) or normal rabbit IgG (closed symbols). (c) TR-BBB13 cells were cultured with PCT-CM (square symbols) or control medium (diamond symbols) pre-treated with anti-angiopoietin-1 antibody (open symbols) or normal rabbit IgG (closed symbols). The occludin mRNA levels were determined by quantitative real-time PCR analysis. Each column or point represents the mean ± SEM (n = 3). Each mRNA expression level was normalized with respect to the β-actin mRNA expression. *p < 0.05, significantly different from the control.

Induction of occludin mRNA in TR-BBB13 cells by fractionated AST-CMs

To determine the approximate molecular weight of TR-AST4 cell-derived soluble factors regulating the occludin mRNA level, a study using fractionated AST-CMs was performed (Fig. 5). The occludin mRNA level was increased by 2.50-, 1.99- and 2.44-fold following treatment with AST-CM (> 10-kDa CM), 30- to 50-kDa and 50- to 100-kDa AST-CM, respectively (Fig. 5a). In contrast, the JAM mRNA level in TR-BBB13 cells was not changed by any of the fractionated AST-CMs (Fig. 5b).

Figure 5.

Effects of fractionated AST-CMs on the occludin (a) and JAM (b) mRNA levels in TR-BBB13 cells. TR-BBB13 cells were cultured with fractionated AST-CM [> 10 (black column), 10–30, 30–50, 50–100 and > 100 kDa fractionated CM (gray columns)] for 24 h. Open column indicates treatment with control medium (serum-free). The occludin and JAM mRNA levels were determined by quantitative real-time PCR analysis. Each column represents the mean ± SEM (n = 3). Each mRNA expression level was normalized with respect to the β-actin mRNA expression. **p < 0.01, *p < 0.05, significantly different from the control.

Identification of angiopoietin-1 secretion from TR-PCT1 cells

The mRNA expression of angiopoietin-1 was detected in TR-PCT1 cells at the same level as in rat lung, used as a positive control, but was not detected in TR-AST4 cells (Fig. 6a). Western blot analysis was performed for PCT-CM with or without reducing agents in order to clarify whether angiopoietin-1 secreted from TR-PCT1 cells forms a complex via disulfide bonds, as reported previously (Procopio et al. 1999) (Fig. 6b). Several bands at > 250 kDa were detected in PCT-CM under non-reducing conditions. Recombinant human angiopoietin-1, used as a positive control, was detected at > 250 kDa under non-reducing conditions. The reducing conditions led to a reduction in the apparent molecular weight of bands. The shifted band was single, and detected at 65 and 75 kDa in PCT-CM and recombinant human angiopoietin-1, respectively.

Figure 6.

Secretion of angiopoietin-1 from TR-PCT1 cells. (a) RT-PCR analysis of angiopoietin-1 and β-actin mRNA in TR-AST4 cells, TR-PCT1 cells and rat lung as a positive control. *Respective RT(–) for left-hand lane. (b) Western blot analysis of angiopoietin-1 in PCT-CM in the presence (Reduced) or absence (Non-reduced) of reducing agents (dithiothreitol and 2-mercaptoethanol). Recombinant human angiopoietin-1 was used as a positive control.

Expression and tyrosine phosphorylation of Tie-2 in TR-BBB13 cells treated with PCT-CM or angiopoietin-1

Western blot analysis revealed that Tie-2 protein, a tyrosine kinase receptor of angiopoietin-1, was expressed in TR-BBB13 cells and had the same molecular weight as that expressed in rat brain and isolated rat brain capillary, used as positive controls (Fig. 7a). The tyrosine phosphorylation of Tie-2 protein after 24-h treatment with two-fold concentrated PCT-CM was examined by immunoprecipitation and western blot analyses (Figs 7b and c). The immunoprecipitated Tie-2 protein was electrophoresed and immunoblotted with anti-PTyr antibody. As shown in Fig. 7(b) (upper panel), the tyrosine phosphorylation of Tie-2 protein increased in TR-BBB13 cells treated with PCT-CM or 300 ng/mL recombinant human angiopoietin-1 (a positive control) compared with that in non-treated TR-BBB13 cells. In contrast, the phosphorylated Tie-2 protein was unchanged in TR-BBB13 cells treated with AST-CM. The amount of Tie-2 protein was not affected by any of the treatment conditions (Fig. 7b, lower panel). The ratio of tyrosine phosphorylated Tie-2 to total Tie-2 density in TR-BBB13 cells treated with AST-CM, PCT-CM or recombinant human angiopoietin-1 was, respectively, 0.89-, 2.60- or 3.55-fold greater than that in non-treated TR-BBB13 cells (Fig. 7c).

Figure 7.

Expression of Tie-2 (a) and effects of PCT-CM and recombinant human angiopoetin-1 on Tie-2 tyrosine phosphorylation (b and c) in TR-BBB13 cells. (a) Western blot analysis of Tie-2 in rat brain and isolated rat brain capillary as positive controls, and TR-BBB13 cells. (b) TR-BBB13 cells were cultured with two-fold concentrated AST-CM, PCT-CM or 300 ng/mL recombinant human angiopoietin-1. After 24 h, cellular extracts from the treated cells were immunoprecipitated (IP) with anti-Tie-2 antibody followed by western blot analysis (WB) with anti-PTyr antibody (upper) or anti-Tie-2 antibody (lower). A typical result from five experiments is shown. (c) The ratio of tyrosine phosphorylated Tie-2 density to total Tie-2 density. Ang-1, recombinant human angiopoietin-1. Each column represents the mean ± SEM (n = 5). **p < 0.01, significantly different from non-treated cells (control).

Discussion

The present study demonstrated that soluble factors secreted from TR-AST4 cells and TR-PCT1 cells induced occludin expression in TR-BBB13 cells. Angiopoietin-1 in PCT-CM predominantly induced occludin expression via Tie-2 tyrosine phosphorylation in TR-BBB13 cells, and was secreted from TR-PCT1 cells as a multimeric disulfide-linked complex (an active form of angiopoietin-1), but not secreted from TR-AST4 cells. This is the first direct evidence concerning the signaling pathway for occludin induction from brain pericytes to BCECs.

The amounts of occludin mRNA and protein were increased in TR-BBB13 cells in transfilter co-culture with cells of the type II astrocyte cell line, TR-AST4 cells (Fig. 1). Type II astrocytes appear to exist in the mature CNS, corresponding to mature fibrous astrocytes mainly localized in white matter (Miller and Raff 1984). This result suggests that fibrous astrocytes ensheathing microvessels at least play some role in maintaining occludin expression at the BBB, mediated by releasing factors and/or a direct contact effect. The concentration-dependent induction of occludin mRNA by AST-CM revealed that soluble factors secreted from TR-AST4 cells induced the occludin gene expression in TR-BBB13 cells (Fig. 2). Direct contact with TR-AST4 cells was also possibly involved in the occludin induction in TR-BBB13 cells, since the membrane pore size (3.0 µm) of the cell culture insert used in the present study was the same as that used in the report in which direct cell-to-cell contact was demonstrated by electron microscopic analysis (Hayashi et al. 1997). Moreover, PCT-CM increased the occludin mRNA level, as well as AST-CM (Fig. 2), suggesting that soluble factors secreted from both type II astrocytes and brain pericytes contribute to the induction of occludin expression at the BBB.

Following treatment with angiopoietin-1, the occludin mRNA level was increased in TR-BBB13 cells (Fig. 3a). The expression of occludin mRNA tends to increase in the range of 1–500 ng/mL angiopoietin-1 treatment, and its mRNA level of 500 ng/mL was significantly different from that of 1 ng/mL. This trend is compatible with the reported KD value of angiopoietin-1 for Tie-2 (about 3 nm, it is calculated as 173 ng/mL when the molecular weight of angiopoietin-1 is 57.7 kDa) (Maisonpierre et al. 1997). This result raised the possibility that the occludin gene induction by AST-CM and PCT-CM was partly mediated by angiopoietin-1. The inhibition study using anti-angiopoietin-1-neutralizing antibody revealed that angiopoietin-1 is mainly responsible for the occludin-inducing activity produced by PCT-CM, as shown in Fig. 4(c). Moreover, RT-PCR analysis confirmed that TR-PCT1 cells expressed angiopoietin-1 mRNA (Fig. 6a). Therefore, angiopoietin-1 is suggested to be the predominant occludin-inducing factor secreted from brain pericytes.

In contrast, occludin induction by AST-CM was not inhibited by treatment with angiopoietin-1-neutralizing antibody (Fig. 4b), and TR-AST4 cells did not express angiopoietin-1 mRNA (Fig. 6a). These results indicate that the soluble factors inducing occludin mRNA in AST-CM are different from angiopoietin-1. The absence of angiopoietin-1 in TR-AST4 cells reflects that in mature astrocytes; it has been reported that the mRNA expression level of angiopoietin-1 in astrocytes is low after 3 weeks after birth (Acker et al. 2001). During the present study, Lee et al. reported that angiopoietin-1 secreted from astrocytes under reoxygenated conditions induces occludin expression in the developing BBB (Lee et al. 2003). Taken together, these results suggest that pericyte-derived angiopoietin-1 and astrocyte-derived undefined factors act to induce occludin expression in the mature BBB. Moreover, angiopoietin-1 seems to be a crucial factor for occludin induction at both the mature and developing BBB, although the cell types secreting angiopoietin-1 seem to change with aging.

The fractionated AST-CM study revealed that the 30- to 50-kDa and 50- to 100-kDa fractions contain factors involved in inducing occludin expression (Fig. 5), although these factors remained undefined in the present study. Interleukin-15 and prolactin have been reported to induce occludin expression in epithelial cells (Stelwagen et al. 1999; Nishiyama et al. 2001). However, the size of the secreted interleukin-15 and prolactin is less than 20 kDa. Accordingly, unknown factors other than interleukin-15 and prolactin induce occludin expression in TR-AST4 cells. Identification of the occludin-inducing factors using AST-CM should provide a better understanding of the physiological regulators of expression of occludin secreted from type II astrocytes.

VEGF and TGF-β1 are known to be increased in the brain in various neurodegenerative diseases (Kalaria et al. 1998; Lesne et al. 2002). Following treatment with VEGF, the occludin mRNA level was reduced in a concentration-dependent manner in TR-BBB13 cells (Fig. 3a). This result is consistent with a previous report on primary bovine BCECs (Wang et al. 2001). TGF-β1 also reduced the occludin mRNA level in a concentration-dependent manner in TR-BBB13 cells (Fig. 3a). This suggests that the BBB disruption induced by TGF-β1 occurs at least partly by a reduction in occludin. The TGF-β1 mRNA level was enhanced in TR-AST4 cells during a 24-h period of hypoxia (Fig. 3b). This was the case for the in vivo response of astrocytes under ischemic conditions (Knuckey et al. 1996), suggesting that type II astrocytes may reduce occludin expression in BCECs under ischemic conditions by producing TGF-β1. In the light of these findings, the occludin expression is regulated by astrocyte- and pericyte-derived factors under physiological and pathophysiological conditions (Fig. 8).

Figure 8.

Postulated mechanism of occludin gene regulation by astrocytes and pericytes for maintaining the adult BBB. [1], Kalaria et al. (1998); [2], Heinsen and Heinsen (1983); [3], Verbeek et al. (1997); [4], Maisonpierre et al. (1997).

JAM expression was unchanged by TR-AST4 cells and TR-PCT1 cells (Figs 1 and 2). This suggests that regulation of JAM expression at the BBB is not regulated by type II astrocytes and brain pericytes, unlike that of occludin. JAM plays a role in inflammatory transmigration of leukocytes as a ligand of integrins (Ostermann et al. 2002). The function of JAM in leukocyte migration may not be influenced by microenvironmental stimuli such as astrocytes and pericytes, VEGF and TGF-β1. Although the concentrated-CM tends to reduce JAM mRNA levels in the left panel of Fig. 2(b), it is not significantly different from non-concentrated-CM. The concentrated-CM would not affect the JAM mRNA in TR-BBB13 cells because other concentrated-CMs did not change JAM or occludin mRNA levels from one- to five-fold concentrated-CM (Fig. 2).

Induction of occludin expression in brain microvessels possibly enhances the tightness of TJs, since overexpression of occludin in MDCK cells has been reported to elevate their TEER (McCarthy et al. 1996). However, the TEER in TR-BBB13 cells was unchanged by the induction of occludin expression following the transfilter co-culture. Claudin-5 is involved in TJ formation at the BBB, and the BBB of the corresponding knock-out mouse showed a loss of tightness for solutes with a permeability under 800 Da (Nitta et al. 2003). Therefore, it is conceivable that TJ formation at the BBB is necessary for claudin-5 as well as occludin. Indeed, the expression level of claudin-5 mRNA in TR-BBB13 cells was very low compared with that of occludin mRNA (data not shown).

The multimeric character of the angiopoietin-1 secreted from pericytes was demonstrated using TR-PCT1 cells (Fig. 6). The 65-kDa protein in PCT-CM is possibly monomeric angiopoietin-1 since it is the same size as recombinant rat angiopoietin-1 reported previously (Iizasa et al. 2002). The size-difference between PCT-CM (65 kDa) and recombinant human angiopoietin-1 (75 kDa) is possibly due to the degree of glycosylation exhibited by difference in species and host cells. The low mobility of the bands detected under non-reducing conditions indicates that angiopoietin-1 forms a disulfide-linked complex, mostly larger than tetrameric. Tetrameric angiopoietin-1 has been recently reported to be the minimal size required for Tie-2 activation (Davis et al. 2003). Therefore, a multimeric complex of angiopoietin-1 endogenously secreted from TR-PCT1 cells is very likely to activate the Tie-2 receptor.

Tyrosine phosphorylation of Tie-2 in TR-BBB13 cells was induced by PCT-CM as shown in Fig. 7. Therefore, the signaling pathway from angiopoietin-1 multimers in PCT-CM to occludin mRNA induction through Tie-2 autophosphorylation is clearly demonstrated. The p85 subunit of phosphatidylinositol 3-kinase (PI3-kinase), the adaptor proteins Grb2, Grb7 and Grb14, and the protein tyrosine phosphatase Shp2 can interact with Tie-2 in a phosphotyrosine-dependent manner (Jones et al. 1999), and are potential downstream molecules of Tie-2. Among these molecules, PI3-kinase has been reported to influence TJ function although this is controversial. It has been reported that PI3-kinase activation increases the amount of occludin localized at the cell surface (Little et al. 2003) although this, conversely, leads to TJ disruption (Woo et al. 1999). Further studies are needed to elucidate the signal molecules involved in occludin induction following the angiopoietin-1/Tie-2 signal.

The present study suggests that the attenuation or inhibition of the angiopoitin-1/Tie-2 pathway between BCECs and brain pericytes leads to dysfunction of the TJs at the BBB in disease. There are two possible ways of modifying the pathway; one is changing the secretion of angiopoietin-1 from pericytes, and the other is inhibition of the interaction between angiopoietin-1 and Tie-2 (Fig. 8). As far as the latter is concerned, angiopoietin-2 is known to suppress angiopoietin-1 activity as an antagonist of Tie-2 (Maisonpierre et al. 1997). In PCT-CM, angiopoietin-1 is possibly a dominant factor compared with angiopoietin-2, since angiopoietin-1-dependent occludin induction is seen following treatment with PCT-CM (Fig. 4c). The expression level of angiopoietin-2 is induced under hypoxic conditions in brain, whereas angiopoietin-1 expression is unchanged (Mandriota et al. 2000). Therefore, it is conceivable that angiopoietin-2 affects occludin expression under pathophysiological conditions by modifying the angiopoietin-1/Tie-2 pathway. The occludin regulation by angiopoietin-1 and angiopoietin-2 is very important for a better understanding of the regulation of TJs at the BBB. Mouse angiopoietin-3 and human angiopoietin-4, other members of the angiopoietin family, are also ligands of Tie-2 (Valenzuela et al. 1999). There is little possibility that angiopoietin-3 and angiopoietin-4 modulate the TJ function at the BBB because their expression was not detected in the brain (Valenzuela et al. 1999), whereas the contribution of angiopoietin-3 and angiopoietin-4 to occludin regulation is still unknown.

Pericyte degeneration seen with aging (Heinsen and Heinsen 1983; Peters et al. 1991) and in neurodegenerative conditions (Verbeek et al. 1997) is associated with increased permeability of the BBB. Moreover, it has been suggested that the reduction of the occludin content at the rat BBB takes place in an age-dependent manner (Mooradian et al. 2003). The findings of the present study indicate that one of the mechanisms of occludin reduction would be a decrease in angiopoietin-1 from pericytes, and this may partly explain the increased permeability following pericyte loss (Fig. 8).

In conclusion, in vitro BBB model studies have demonstrated that soluble factors (the molecular weight range of candidates is 30–100 kDa) secreted from type II astrocytes and multimeric angiopoietin-1, activating the Tie-2 receptor via tyrosine phosphorylation, secreted from pericytes induce occludin expression. In contrast, VEGF and TGF-β1, the mRNA of which was up-regulated in type II astrocytes following hypoxic treatment, reduced occludin mRNA. These findings provide important information about the molecular mechanism of TJ regulation based on paracrine interactions between BECEs, and astrocytes and pericytes under both physiological and pathophysiological conditions, and should help in the design of molecules to block TJ disruption and, thereby, protect the CNS.

Acknowledgements

The authors wish to thank Drs K. Tetsuka, T. Asashima and Messrs. H. Iizasa, T. Kondo for valuable discussions and Ms. N. Funayama for secretarial assistance. This study was supported, in part, by a Grant-in-Aid for Scientific Research, and a 21st Century Center of Excellence (COE) Program from Japan Society for the Promotion of Science. It was also supported in part by the Tokyo Biochemical Research Foundation and the Industrial Technology Research Grant Program in ‘00 from the New Energy and the Industrial Technology Development Organization (NEDO) of Japan.

Ancillary