Internalization of basic fibroblast growth factor at the mouse blood–brain barrier involves perlecan, a heparan sulfate proteoglycan

Authors

  • Yoshiharu Deguchi,

    1. Department of Drug Disposition and Pharmacokinetics, School of Pharmaceutical Sciences, Teikyo University, Kanagawa, Japan
    Search for more papers by this author
  • Hiroshi Okutsu,

    1. Department of Biopharmacy, School of Pharmaceutical Sciences, University of Shizuoka, Shizuoka, Japan
    Search for more papers by this author
  • Takashi Okura,

    1. Department of Biopharmacy, School of Pharmaceutical Sciences, University of Shizuoka, Shizuoka, Japan
    Search for more papers by this author
  • Shizuo Yamada,

    1. Department of Biopharmacy, School of Pharmaceutical Sciences, University of Shizuoka, Shizuoka, Japan
    Search for more papers by this author
  • Ryohei Kimura,

    1. Department of Biopharmacy, School of Pharmaceutical Sciences, University of Shizuoka, Shizuoka, Japan
    Search for more papers by this author
  • Takuro Yuge,

    1. FGF Product Team, Kaken Pharmaceutical Co., Ltd, Tokyo, Japan
    Search for more papers by this author
  • Akihiko Furukawa,

    1. FGF Product Team, Kaken Pharmaceutical Co., Ltd, Tokyo, Japan
    Search for more papers by this author
  • Kazuhiro Morimoto,

    1. Department of Biopharmaceutics, Hokkaido College of Pharmacy, Hokkaido, Japan
    Search for more papers by this author
  • Masanori Tachikawa,

    1. Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan
    Search for more papers by this author
  • Sumio Ohtsuki,

    1. CREST of Japan Science and Technology Corporation, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan
    2. Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan
    Search for more papers by this author
  • Ken-ichi Hosoya,

    1. CREST of Japan Science and Technology Corporation, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan
    2. Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan
    Search for more papers by this author
  • Tetsuya Terasaki

    1. CREST of Japan Science and Technology Corporation, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan
    2. Department of Molecular Biopharmacy and Genetics, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan
    Search for more papers by this author

Address correspondence and reprint requests to Yoshiharu Deguchi, Department of Drug Disposition and Pharmacokinetics, School of Pharmaceutical Sciences, Teikyo University, 1091–1 Suarashi, Sagamiko-machi, Tsukui-gun, Kanagawa, 199–0195, Japan. E-mail: deguchi@pharm.teikyo-u.ac.jp

Abstract

In this study, the internalization mechanism of basic fibroblast growth factor (bFGF) at the blood–brain barrier (BBB) was investigated using a conditionally immortalized mouse brain capillary endothelial cell line (TM-BBB4 cells) as an in vitro model of the BBB and the corresponding receptor was identified using immunohistochemical analysis. The heparin-resistant binding of [125I]bFGF to TM-BBB4 cells was found to be time-, temperature-, osmolarity- and concentration-dependent. Kinetic analysis of the cell-surface binding of [125I]bFGF to TM-BBB4 cells revealed saturable binding with a half-saturation constant of 76 ± 24 nm and a maximal binding capacity of 183 ± 17 pmol/mg protein. The heparin-resistant binding of [125I]bFGF to TM-BBB4 was significantly inhibited by a cationic polypeptide poly-L-lysine (300 µm), and compounds which contain a sulfate moiety, e.g. heparin and chondroitin sulfate-B (each 10 µg/mL). Moreover, the heparin-resistant binding of [125I]bFGF in TM-BBB4 cells was significantly reduced by 50% following treatment with sodium chlorate, suggesting the loss of perlecan (a core protein of heparan sulfate proteoglycan, HSPG) from the extracellular matrix of the cells. This type of binding is consistent with the involvement HSPG-mediated endocytosis. RT-PCR analysis revealed that HSPG mRNA and FGFR1 and FGFR2 (tyrosine-kinase receptors for bFGF) mRNA are expressed in TM-BBB4 cells. Moreover, immunohistochemical analysis demonstrated that perlecan is expressed on the abluminal membrane of the mouse brain capillary. These results suggest that bFGF is internalized via HSPG, which is expressed on the abluminal membrane of the BBB. HSPG at the BBB may play a role in maintaining the BBB function due to acceptance of the bFGF secreted from astrocytes.

Abbreviations used
BBB

blood–brain barrier

bECGF

bovine endothelial cell growth factor

bFGF

basic fibroblast growth factor

DMEM

Dulbecco's modified Eagle's medium

FBS

fetal bovine serum

FGFR

fibroblast growth factor receptor

HBSS

Hank's balanced salt solution

HSPG

heparan sulfate proteoglycans

PBS

phosphate-buffered saline

TCA

trichloroacetic acid

TFA

trifluoroacetic acid.

The blood–brain barrier (BBB), which is formed by complex tight-junctions of the brain capillary endothelial cells, is an important regulatory system for maintaining the neuronal activity of the CNS. One of the most important functions of the BBB is to regulate the transport of nutrients, hormones and drugs between the circulating blood and the brain interstitial fluid. Although the mechanism for maintaining BBB function has not been fully clarified yet, one of the proposed mechanisms involves paracrine interaction between brain capillary endothelial cells and astrocytes (Hayashi et al. 1997). Recently, Sobue et al. (1999) suggested that basic fibroblast growth factor (bFGF) is one of the most plausible soluble factors secreted from astrocytes to enhance the barrier properties of the BBB.

bFGF is a 18-kDa polypeptide composed of 154 amino acid residues with an isoelectric point of 10.1. This cationic peptide, which is highly expressed in the CNS, may play an important role in regeneration after injury of the CNS and participate in a cascade of neurotrophic events facilitating neuronal repair and survival (Morrison et al. 1986; Finklestein et al. 1993; MacMillan et al. 1993; Endoh et al. 1994; Bikfalvi et al. 1997). The immunolocalization of bFGF was observed in astrocytes and neuronal cells, suggesting that these cells may produce bFGF and act as neurotrophic factors (Bikfalvi et al. 1997). Moreover, bFGF is internalized in cells via heparan sulfate proteoglycans (HSPG) and/or tyrosine kinase receptors FGFRs (FGFR1 and FGFR2) to regulate the intracellular biological response in a variety of cells (Roghani and Moscatelli 1992; Rusnati et al. 1993; Gleizes et al. 1995; Sperinde and Nugent 1998). If bFGF acts as a trophic factor in the brain capillary endothelial cells to regulate BBB function as well as in glial and neuronal cells, this will involve binding to the receptors localized on the abluminal membrane of the brain capillary and internalization into the cells. However, our knowledge about the internalization mechanism of bFGF into brain capillary endothelial cells and the identification of the corresponding receptors is still incomplete.

The purpose of this study was to investigate the mechanism of internalization of bFGF into a conditionally immortalized mouse brain capillary endothelial cell line, TM-BBB4, as an in vitro model of the BBB (Hosoya et al. 2000; Terasaki and Hosoya 2001). Moreover, HSPG core proteins and/or FGFRs mRNA expression in TM-BBB4 and HSPG protein expression at the mouse brain were examined by reverse transcription-polymerase chain reaction (RT-PCR) and immunohistochemical analysis, respectively.

Materials and methods

Materials

Recombinant human basic fibroblast growth factor (154 amino acid residues, 17 kDa, pI = 10.1) was supplied by Kaken Pharmaceutical Co., Ltd (Tokyo, Japan). Iodinated bFGF [125I]bFGF: 20.8 MBq/mL was prepared by the modified lactoperoxidase method (Yuge et al. 1997). The radiochemical purity of [125I]bFGF was confirmed to be 98% or more by HPLC using a TSKgel G2000 SWXL gel filtration column (Tosoh, Tokyo, Japan). Rat anti-HSPG monoclonal antibody was purchased from Chemicon International Inc. (Temecula, CA, USA). All other chemicals were of analytical grade and were used without further purification.

Cell culture

TM-BBB4 cells were grown routinely in collagen type 1-coated 75 cm2 tissue flasks (BD Biosciences, MA, USA) at 33°C under 5% CO2/air. The permissive-temperature of the TM-BBB4 cell culture is 33°C because of the expression of temperature-sensitive large T-antigen (Hosoya et al. 2000). The culture medium was Dulbecco's modified Eagle's medium (DMEM, Nissui Pharmaceutical Co., Ltd, Tokyo, Japan) supplemented with 1.5 mg/mL sodium bicarbonate, 15 µg/mL bovine endothelial cell growth factor (bECGF, Roche Molecular Biochemicals, Mannheim, Germany), 70 µg/mL benzylpenicillin potassium, 100 µg/mL streptomycin sulfate, and 10% fetal bovine serum (FBS, Moregate, Bulimba, Australia).

Pulse-chase investigation of cell-binding and internalization of [125I]bFGF

TM-BBB4 cells were seeded at a density of 3–5 × 104 cells/cm2 on collagen type 1-coated 24-well plates (BD Biosciences) with DMEM-based cell culture medium as described above, and cultured for 72–96 h. Subconfluent cultures of TM-BBB4 cells were washed twice with cold PBS (pH 7.4) and preincubated for 20 min at 4°C, and then for 20 min in modified Hank's balanced salt solution (HBSS, 138 mm NaCl, 1.3 mm CaCl2, 5.0 mm KCl, 0.8 mm MgCl2, 0.3 mm KH2PO4, 0.3 mm Na2HPO4, 5.6 mm d-glucose, plus 10 mm HEPES, 0.15% gelatin and 0.1% bovine serum albumin pH 7.4) containing [125I]bFGF (330 pm∼ 3 µm) and the agents to be tested (pulse period). After the pulse period, which allowed [125I]bFGF to reach equilibrium with its cell-surface binding sites, the incubation medium was changed to modified HBSS without [125I]bFGF. Then, the temperature of the cultures was changed to 37°C for different periods of time (chase period) to promote internalization of cell-surface bound [125I]bFGF by the cells. After incubation, the cells were washed three times with cold PBS (pH 7.4), followed by 2 m NaCl containing 100 µg/mL heparin for 10 min to remove cell-surface bound radioactivity. The cultures were then used to determine the total binding (cell-surface binding plus internalization) and heparin-resistant binding (internalization) of [125I]bFGF using a γ-counter after the cells were lysed with 1 m NaOH. The protein content was determined by a BCA protein assay kit (Pierce, Chemical Co., Rockford, IL, USA).

Treatment of TM-BBB4 cells with hypertonic medium, chlorate and heparinase-I

In order to confirm the internalization of [125I]bFGF by TM-BBB4 cells, the cells were incubated with a hypertonic buffer (the modified HBSS plus 1.2 m sucrose, 1500 mOsm/kg) containing [125I]bFGF (5.3 kBq/mL, 31.2–48.6 nm) for 20 min at 4°C. Then, the chase experiments were carried out at 37°C for the time designated.

Subconfluent TM-BBB4 cells were cultured with DMEM-based culture medium containing 10% FBS and 50 mm sodium chlorate for 48 h before the start of the pulse-chase experiments. For heparinase-I treatment, the cells were incubated with the modified HBSS containing 10 units/mL heparinase-I for 2 h at 37°C, and then the pulse-chase experiments were carried out as described above.

RT-PCR analysis

Total RNA was prepared from PBS-washed TM-BBB4 cells using ISOGEN (Nippon Gene, Tokyo, Japan) according to the protocol. Reverse transcription and PCR amplification were carried out with GeneAmp (PCR system 9700, Perkin-Elmer, Norwalk, CT, USA). Single-strand cDNA was synthesized from 1 µg total RNA by reverse transcription (RevaTra Ace, Toyobo, Osaka, Japan) using an oligo-dT primer. The primers used for amplification had the following sequences: FGFR1 (sense, 5′-TTCTGGGCTGTGCTGGTAC-3′ and antisense, 5′-GCGAACCTTGTAGCCTCCAA-3′), FGFR2 (sense 5′-TTCATCTGCCTGGTCTTGGT-3′ and antisense, 5′-AATAAGGCTCCAGTGCTGGTTTC-3′), perlecan (sense 5′-CAGGTCCTAATGTGGCGGTCAACAC-3′ and antisense 5′-CAATCCCCTTGTCCTGCCCAT-3′), syndecan2 (sense 5′-ACAGCCTGGCTGCTTAAGATGGATGT-3′ and antisense 5′-TGGAAGGGCCCATCATTTCCTTTTA-3′), syndecan3 (sense 5′-CAGCTACCTCTCGGCCACAATCC-3′ and antisense 5′-CACCTCCTTCCGCTCTAGTATGCTCTT-3′). PCR was performed using Hot Star Taq DNA polymerase (QIAGEN, Hilden, Germany) with the following thermal cycle: 1 cycle at 95°C for 15 min, 35 cycles of 94°C for 30 s, 60°C for 30 s, 72°C for 1 min, and 1 cycle of 72°C for 10 min. The PCR products were separated by electrophoresis in 2.5% agarose gel and visualized with an imager (EPIPRO 7000, Aisin, Aichi, Japan) in the presence of ethidium bromide (10 mg/mL). The amplified products were then subcloned into a pGEM-T Easy Vector (Promega, Madison, WI, USA) and sequenced by a DNA sequencer (model 4200, Li-COR, Lincoln, NE, USA).

Immunohistochemical studies in TM-BBB4 cells and mouse brain

TM-BBB4 cells were seeded at a density of 3–4 × 103cells on collagen type I-coated cover glass (Asahi Techno Glass Corp., Tokyo, Japan) with DMEM-based cell culture medium, and cultured for 24 h. TM-BBB4 cells were washed with phosphate-buffered saline (PBS) and fixed with 3.7% formaldehyde for 30 min at room temperature. Cells were permeated with 0.2% Triton X-100 in PBS for 15 min and incubated with 10% goat serum for 20 min. After washing with PBS, cells were further incubated with rat antiperlecan antibody (1 : 100 dilution) (Chemicon International Inc.) as a primary antibody with 1.5% goat serum in PBS for 1 h at room temperature. Cells were then washed three times with PBS and incubated for 45 min at room temperature with 10 µg/mL rhodamine conjugated goat anti-rat IgG (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA). After extensive washing, the cover glass was mounted in 90% glycerol on a microscope slide. Cells were then viewed by confocal laser scanning microscopy (TCS SP, Leica, Heidelberg, Germany). The control experiments were carried out in parallel using normal rat IgG, instead of the primary antibody.

Adult male ddY mouse were anesthetized by intramuscular injection of ketamine. Adult mouse brains were perfused transcranially with 4% paraformaldehyde in 0.1 m sodium phosphate buffer (pH 7.2). Cryostat sections (30 µm in thickness; CM1900, Leica) were prepared for immunofluorescence. All immunohistochemical incubations were performed at room temperature. Sections were incubated with rat anti-HSPG antibody overnight (1 : 250 dilution), and then incubated with FITC-conjugated secondary antibody (ICN, OH) for 2 h. For double immunostaining, brain cryosections were incubated with anti-HSPG antibody (1 : 250 dilution) and anti-GFAP antibody (1 : 1 dilution, Dako, Palo Alto, CA, USA). The immunofluorescent-stained sections were viewed using a confocal laser scanning microscope.

HPLC method

[125I]bFGF and its metabolites were separated by HPLC from the cells lysed with 1% Triton X-100 (Nacalai Tesque Inc., Tokyo, Japan) at 4°C for 60 min The lysate was lyophilized (FD-80, EYELA, Tokyo, Japan), and then reconstituted in mobile phase [2% acetonitrile in 0.1% trifluoroacetic acid (TFA)]. An aliquot of the sample was then subjected to HPLC using the following system: HPLC column, Protein C18, (250-mm × 4.6-mm, Vydac, The Separation Group Inc., Hesperia, CA, USA); pump 880-PU (Jasco, Tokyo, Japan); mobile phase mixer 880–30 (Jasco). The mobile phase was 2–50% acetonitrile in 0.1% TFA (linear gradient system).

Data analysis

The data for total and heparin-resistant binding were expressed as the cell-to-medium concentration ratio.

image

The cell-surface binding data were analyzed by a model that involved saturable and non-saturable binding as follows:

image

where Bmax and Kd are the maximal binding capacity and the dissociation constant for binding, respectively. An α represents a non-saturable binding constant. Cf is the free concentration of unchanged [125I]bFGF in the incubation medium, corrected by trichloroacetic acid (TCA) precipitability.

Unless otherwise indicated, all data represent the mean ± SE. Statistical significance among means of more than two groups was determined by one way analysis variance followed by the modified Fisher's least squares difference method.

Results

Total and heparin-resistant binding of [125I]bFGF in TM-BBB4 cells

Figure 1 shows the time-courses of the total and heparin-resistant binding of [125I]bFGF at 37°C in the pulse-chase experiment. The total binding rapidly reached equilibrium with a cell/medium ratio of 250 µL/mg protein. On the other hand, the heparin-resistant binding increased with time up to 60 min. The cell/medium ratio at equilibrium reached 60 µL/mg protein, indicating that [125I]bFGF is internalized with time, translocating from the surface to the inside of the cell.

Figure 1.

Time-courses of the total and heparin-resistant binding of [125I]bFGF to TM-BBB4 by the pulse–chase method. The cells were incubated with [125I]bFGF (31.2 nm) on ice for 20 min. Then, the cells were washed with PBS, and incubated with incubation buffer without [125I]bFGF at 37°C for the predetermined time period. At the end of incubation, the cells were washed three times with cold PBS (total binding), then washed with 2 m NaCl containing 100 mg/mL heparin for 10 min (heparin-resistant binding). Total binding (●); heparin-resistant binding (○).

Figure 2 shows the effects of temperature and medium osmolarity on the heparin-resistant binding of [125I]bFGF to TM-BBB4 cells. No time-dependent increase was observed in the heparin-resistant binding of [125I]bFGF at 4°C. The cell/medium ratio at 4°C at 60 min was significantly reduced compared with the value at 37°C. Treatment of TM-BBB4 cells with hypertonic medium (1500 mOsm/kg) showed a marked reduction in the heparin-resistant binding of [125I]bFGF due to a decrease in the intracellular water space of TM-BBB4 cells (Fig. 2).

Figure 2.

Effect of temperature, increased osmolarity of the incubation medium on the heparin-resistant binding of [125I]bFGF to TM-BBB4. The cells were incubated with [125I]bFGF (31.2–34.7 nm) at 4°C for 20 min. Then, the cells were washed with PBS, and incubated with the incubation buffer without [125I]bFGF at 37°C or 4°C. To examine the effect of the medium osmolarity, the cells were incubated with hypertonic medium (1500 mOsm/kg) including [125I]bFGF (31.2–34.7 nm) at 4°C for 20 min. After the end of the pulse period, cells were washed with PBS, then incubated with the modified HBSS without the labeled peptide at 37°C for the predetermined time period. Significantly different from control, *p < 0.05. 37°C (●); 4°C (○); hypertonic medium (1500 mOsm/kg) (□).

The metabolism of [125I]bFGF during the course of the pulse-chase experiments was examined by HPLC. As shown in Fig. 3, the HPLC chromatogram of authentic [125I]bFGF had a single peak at a retention time of 43 min. On the other hand, an HPLC chromatogram of [125I]bFGF extracted from TM-BBB4 cells showed double peaks at retention times of 40 and 43 min. Unchanged [125I]bFGF accounted for 25% of the total radioactivity.

Figure 3.

An HPLC chromatogram of [125I]bFGF internalized by TM-BBB4 cells. [125I]bFGF internalized by TM-BBB4 cells using the pulse-chase method was extracted by solubilizing the cells and subjected to HPLC. An HPLC chromatogram of a standard sample of [125I]bFGF is illustrated in the inset.

Concentration-dependence and inhibitory effects of several compounds on the binding of [125I]bFGF in TM-BBB4 cells

The cell-surface binding of [125I]bFGF to TM-BBB4 cells was measured by incubating the cells at 4°C for 120 min. As shown in Fig. 4(a), the cell-surface binding of [125I]bFGF was concentration-dependent over the range 1 nm to 10 µm. Nonlinear regression analysis of the data gave a Bmax of 183 ± 17 pmol/mg protein, Kd of 76 ± 24 nm, and α of 0.046 ± 0.003 mL/mg protein. The heparin-resistant binding of [125I]bFGF to the cells was also examined over the concentration range 200 pm to 1.6 µm. The heparin-resistant binding was also concentration-dependent in parallel with the total binding to the cell surface (solid line in Fig. 4b), supporting the hypothesis that the cell surface binding is a limiting factor for heparin-resistant binding.

Figure 4.

Concentration-dependence of the total (a) and heparin-resistant (b) binding of [125I]bFGF to TM-BBB4. (a) The cells were incubated with [125I]bFGF (1–10 µm) at 4°C for 120 min. Then, the cells were washed with cold PBS and solubilized with 1 m NaOH. (b) After the pulse-chase experiments with [125I]bFGF (200 pm−1.6 µm), the heparin-resistant binding of [125I]bFGF was determined by solubilization of the cells.

Table 1 shows the effects of several compounds on the total and heparin-resistant binding of [125I]bFGF to TM-BBB4 cells. The total binding was inhibited by 97% by 300 µm poly-l-lysine and also inhibited by up to 43% by heparin, chondroitin sulfate-B and fucoidan (10 µg/mL each) which contain a sulfate moiety. On the other hand, hyaluronic acid and dextran (10 µg/mL each) had no effect. The heparin-resistant binding of [125I]bFGF was significantly inhibited by poly-l-lysine, heparin and chondroitin sulfate-B by 95%, 68% and 20%, respectively.

Table 1.  Effects of several compounds on the total and heparin-resistant binding of [125I]bFGF in TM-BBB4 cells
InhibitorConcentrationTotal
binding
(% of control)
Heparin-resistant
binding
(% of control)
  1. Values are mean ±SE of three experiments. The cells were incubated with [125I]bFGF (25.5–33.0 nm) and each compound at 4°C for 20 min. Then, the incubation medium was removed and the cells were incubated with [125I]bFGF-free medium at 37°C for 60 min in the absence (control) or presence of each compound. CS-B, chondroitin sulfate-B; HA, hyaluronic acid. n.d. not determined. Significantly different from control, *p < 0.05, **p < 0.01.

Poly-l-lysine300 µm  3.0 ± 0.6**  5.3 ± 0.3*
Insulin 10 µmn.d.106.7 ± 4.9
Transferrin 10 µmn.d.113.3 ± 1.8
Heparin 10 µg/mL 14.2 ± 1.3** 31.6 ± 0.5**
CS-B 10 µg/mL 66.6 ± 3.9* 79.5 ± 4.6*
Fucoidan 10 µg/mL 59.5 ± 6.4* 90.3 ± 3.6
HA 10 µg/mL 94.6 ± 3.2123.5 ± 10.7
Dextran 10 µg/mL108.8 ± 8.0124.8 ± 11.6

The heparin-resistant binding of [125I]bFGF was examined in TM-BBB4 cells treated with sodium chlorate and heparinase-I. Such treatment removes HSPG from the extracellular matrix of the cells. As shown in Fig. 5, 48-h treatment of cells with sodium chlorate and 2 h treatment with heparinase-I significantly inhibited the heparin-resistant binding of [125I]bFGF by 50% compared with the control value.

Figure 5.

Effects of sodium chlorate, heparinase-I treatment on the heparin-resistant binding of [125I]bFGF in TM-BBB4. After the cells were treated with either 50 mm sodium chlorate for 48 h or 10 units/mL heparinase-I for 2 h, the cells were incubated with [125I]bFGF (17.8–46.7 nm) at 4°C for 20 min. Then, the cells were washed with PBS, and incubated with [125I]bFGF-free HBSS at 37°C for 60 min. Significantly different from control, **p < 0.01.

Expression of HSPG core protein and FGFRs mRNAs in TM-BBB4 cells

RT-PCR analysis was performed to examine the expression of HSPG core proteins and FGFRs mRNAs in TM-BBB4 cells using total RNA isolated from the cells. As shown in Fig. 6(a), HSPG core protein, perlecan at 430 bp, was amplified in TM-BBB4 cells. However, no appreciable products of syndecan2 and syndecan3 were detected. The nucleotide sequence of the band was almost identical to mouse perlecan with a homology of 97% (Gene-Bank accession number: M77174).

Figure 6.

mRNA expression of HSPG core proteins and FGFR family. (a) RT-PCR analysis of HSPG core proteins in TM-BBB4. Lanes 1 and 4, perlecan, lane 2, syndecan2; lane 3, syndecan3. Only perlecan was amplified. (b) RT-PCR analysis of FGFR family mRNA in TM-BBB4. Lane 1, FGFR1; lane 2, FGFR2. RT-PCR was conducted with (+) or without (–) reverse transcription. M, molecular weight markers. The sizes of molecular weight markers are shown on the left.

As far as FGFRs were concerned, FGFR1 at 330 bp and 600 bp, and FGFR2 at 280 bp, 380 bp and 630 bp were amplified in TM-BBB4 cells (Fig. 6b). The nucleotides of these bands were virtually identical to mouse FGFR1 and FGFR2 with a homology of 98% and 100%, respectively (accession numbers: M33760, M63503). Our results are consistent with the results of RT-PCR using total RNA isolated from mouse brain tissue reported by Ozawa et al. (1996).

Immunostaining of HSPG by the mouse brain and TM-BBB4 cells

The localization of HSPG was determined in the cerebral cortex by immunohistochemical analysis (Figs 7a–c). HSPG immunoreactivity was strongly detected in brain capillaries, which are ramified in all cortical layers (Fig. 7a). Immunostaining was observed over the surface of the capillaries (Fig. 7b) and was localized to the outer side of capillary endothelium nuclei (Figs 7b and c). Double immunostaining of HSPG (Fig. 7d) and GFAP (Fig. 7e), which is an astrocyte marker, showed that HSPG immunoreactivity does not completely overlap with that of GFAP (Fig. 7f), supporting the hypothesis that HSPG is expressed at least on the abluminal membrane of the brain capillary and the overlapping signal (yellow) shows contact between the astrocyte foot process and the abluminal membrane of the brain capillary. This characteristic immunostaining was not seen following the use of preimmune rabbit immunoglobulin (data not shown). These features indicate that HSPG is highly expressed on the abluminal membrane of the mouse brain capillary.

Figure 7.

Immunostaining of HSPG in the cerebral cortex of the adult mouse brain and TMBBB4 cells. (a–c) Immunofluorescence of HSPG (green) at low (a) and high magnification (b) of the cerebral cortex, and cross-sectional views of the brain capillary (c). Nuclei were stained by propidium iodide (red). The white arrow head indicates a nucleus of brain capillary endothelial cells. (d–f) Immunofluorescence for HSPG (green; d) and GFAP (red; e), and (f) merged image of (d) and (e). (g) Immunostaining of HSPG (perlecan) by TM-BBB4 cells.

Figure 7(g) shows the confocal microscope image of TM-BBB4 cells stained with antiperlecan antibody. Intense immunoreactivity was detected in TM-BBB4 cells. Such characteristic immunostainings did not appear when rabbit normal IgG was used (data not shown).

Discussion

In the present study, HSPG is expressed in TM-BBB4 cells and mouse brain capillaries and plays a role in the internalization of bFGF into TM-BBB4 cells used as an in vitro model of the BBB.

The heparin-resistant binding gradually increased with time and reached equilibrium after a 60-min incubation period (Fig. 1). The HPLC chromatogram of [125I]bFGF in the heparin-resistant fraction demonstrated that about 25% of the total radioactivity remained as an intact peptide (Fig. 3). The peak with a retention time of 40 min seems to be due to truncated metabolites with molecular weights comparable with the intact [125I]bFGF. This is supported by previous findings that bFGF was significantly metabolized to a TCA-precipitable metabolite (14–16 kDa) in vascular smooth muscle cells (Sperinde and Nugent 2000), liver and kidney after systemic administration (Yuge et al. 1997). These metabolites retain their heparin-binding and mitogenic activities. In addition, the heparin-resistant binding of [125I]bFGF in TM-BBB4 cells was temperature-, osmolarity- and concentration-dependent (Figs 2 and 4b). These results strongly suggest that an intact form of [125I]bFGF is internalized into the cells via endocytosis.

Kinetic analysis of the cell-surface binding of [125I]bFGF to TM-BBB4 cells demonstrated that there was a saturable binding site with a Kd of 76 nm (Fig. 4a), which was similar to the value obtained using bovine brain capillaries (40 nm) (Deguchi et al. 2000) and Chinese hamster ovary cells (Roghani and Moscatelli 1992). However, the Kd was 14–63 times greater than that for RME-transported peptide (Duffy and Pardridge 1987; Fishman et al. 1987; Golden et al. 1997) and 10–200 times smaller than that for AME-transported peptides (Pardridge et al. 1989, 1990; Terasaki et al. 1991; Shimura et al. 1992). From these results, it appears that the endocytosis mechanism of [125I]bFGF at the BBB may be different from that of RME and AME as reported previously.

bFGF consists of a 12 β-strand that possesses two positively charged regions. One region is composed of the residues Ans28, Arg121, Lys126 and Gln135, while the other is composed of the residues Lys27, Asn102 and Lys136 (Nugent and Iozzo 2000). Heparan sulfate consists of a disaccharide repeat unit composed of l-iduronic acid and d-glucosamine that are sulfated and acetylated. The sulfate groups result in a negative cell-surface charge and avidly bind to the positively charged region of bFGF (Turnbull et al. 1992). Therefore, if bFGF can be internalized via the HSPG-mediated pathway into brain capillary endothelial cells, the heparin-resistant binding of [125I]bFGF to TM-BBB4 cells would be inhibited by addition of a cationic polypeptide (poly-l-lysine) and heparan sulfate-like compounds with sulfate groups (heparin and chondroitin sulfate-B) to the incubation medium. Poly-l-lysine could inhibit the binding of [125I]bFGF to HSPG on the cells. In contrast, addition of heparan sulfate-like compounds would reduce unbound [125I]bFGF in the incubation medium. As expected, the total and heparin-resistant binding of [125I]bFGF were inhibited by 300 µm poly-l-lysine, as well as heparin and chondroitin sulfate B (each 10 µg/mL).

Further evidence for the HSPG-mediated endocytosis of bFGF in TM-BBB4 cells was obtained in the experiment using cells devoid of HSPG following treatment with sodium chlorate and heparinase (Gleizes et al. 1995). Heparinase-I used in this study cleaves the α-1,4-d-glycosaminide bond within the HSPG structure (Hovingh and Linker 1970), whereas sodium chlorate inhibits sulfation of the glycosaminoglycans (Fannon and Nugent 1996). The heparin-resistant binding of [125I]bFGF was significantly inhibited by heparinase-I and sodium chlorate by up to 50% (Fig. 5). In addition, RT-PCR and immunohistochemical studies clearly showed that at least perlecan is expressed in TM-BBB4 cells (Figs 6a and 7g). These results strongly suggest that HSPG-mediated endocytosis is involved in the internalization of [125I]bFGF by TM-BBB4 cells. However, the involvement of other internalization process cannot be ruled out at the present time.

Moreover, we have provided the first demonstration that perlecan is highly expressed on the abluminal membrane of the mouse brain capillary (Figs 7a–f). Several types of core proteins of HSPG have been characterized (Iozzo and Murdoch 1996). Among them, perlecan has been suggested to be an important candidate for a bFGF binding site (Aviezer et al. 1994). This suggests that HSPG located on the abluminal membrane of the BBB may be capable of binding and internalizing bFGF. On the other hand, it remains unclear whether the apical side of TM-BBB4 cells (the side which is in contact with the incubation medium) reflects the abluminal side of the brain capillary. However, the present results obtained using TM-BBB4 would, in part, explain the endocytosis from the abluminal side of the brain capillary.

Both FGFR1 and FGFR2, which are tyrosine-kinase receptors and members of the FGFR family encoded by distinct genes, were demonstrated to be expressed in TM-BBB4 cells at the mRNA level (Fig. 6b). FGFR1 and FGFR2 have three immunoglobulin (Ig) domains and an acid box, and they have several splice variants. In the present study, we used the same primers as reported by Ozawa et al. (1996). The area amplified by the primer for FGFR1 includes the first Ig loop of FGFR1, whereas the primer for FGFR2 includes the first Ig loop plus the acid box. Therefore, the plural PCR products in Fig. 6(b) correspond to 597 bp and 383 bp (lacking the first Ig loop) for FGFR1, and 629 bp, 383 bp (lacking the first Ig-loop) and 284 bp (lacking the first Ig-loop and the acid box) for FGFR2.

It has been reported, using cells transfected with cDNA encoding FGFR1 and FGFR2, that bFGF is internalized into the cells via the receptors expressed (Roghani and Moscatelli 1992). Therefore, bFGF would be internalized through the FGFR-mediated pathway which may appear in TM-BBB4 cells. Kinetic analysis of the binding data shown in Fig. 4 failed to provide an estimate of the binding parameters (Kd and Bmax) for the FGFR-mediated pathway. The reason for this may be that the concentration of bFGF in the incubation medium (330 pm∼) exceeded the Kd value for FGFR (20–40 pm) (Roghani and Moscatelli 1992). However, the fact that part of the heparin-resistant binding remained after treatment with sodium chlorate or heparinase-I may indicate the existence of an FGFR-mediated pathway. Detailed studies will be necessary to clarify the contribution of FGFR-mediated internalization of bFGF in TM-BBB4 cells.

It is accepted that astrocytes induce and maintain BBB properties in endothelial cells (Janzer and Raff 1987; Hayashi et al. 1997; Gaillard et al. 2001). There is also a report that bFGF, which is likely to be secreted from astrocytes (Trentin et al. 2001), induces alkaline phosphatase activity (a marker of BBB phenotypes) and reduces l-glucose permeability in immortalized bovine brain endothelial cells (Sobue et al. 1999). Moreover, anti-bFGF antibody abolishes 90% of the activation of alkaline phosphatase activity by astrocyte-conditioned medium, indicating that bFGF is one of the soluble factors necessary for maintaining BBB function. The present studies could clarify part of the cascade where bFGF secreted from astrocytes controls the function of the BBB.

In conclusion, this is the first evidence that perlecan, a core protein of HSPG, is strongly expressed on the abluminal membrane of the mouse brain capillary. HSPG at the BBB mediates bFGF internalization and may be a main receptor for bFGF secreted from astrocytes.

Acknowledgements

This work was supported in part by a Grant-in-Aid for Scientific Research (C) (11672270) provided by the Ministry of Education, Science, Sports and Culture of Japan.

Ancillary