Acta Neuropathol (2002) 104 : 85–91 DOI 10.1007/s00401-002-0524-x
R E G U L A R PA P E R
Gesa Rascher · Arne Fischmann · Stephan Kröger · Frank Duffner · Ernst-H. Grote · Hartwig Wolburg
Extracellular matrix and the blood-brain barrier in glioblastoma multiforme: spatial segregation of tenascin and agrin
Received: 10 July 2001 / Revised: 19 November 2001 / Accepted: 18 December 2001 / Published online: 28 March 2002 © Springer-Verlag 2002
Abstract The quality of the blood-brain barrier (BBB), represented mainly by endothelial tight junctions (TJ), is now believed to be dependent on the brain microenvironment and influenced by the basal lamina of the microvessels. In the highly vascularized glioblastoma multiforme (GBM), a dramatic increase in the permeability of blood vessels is observed but the nature of basal lamina involvement remains to be determined. Agrin, a heparan sulfate proteoglycan, is a component of the basal lamina of BBB microvessels, and growing evidence suggests that it may be important for the maintenance of the BBB. In the present study, we provide first evidence that agrin is absent from basal lamina of tumor vessels if the TJ molecules occludin, claudin-5 and claudin-1 were lacking in the endothelial cells. If agrin was expressed, occludin was always localized at the TJ, claudin-5 was frequently detected, whereas claudin-1 was absent from almost all vessels. Furthermore, despite a high variability of vascular phenotypes, the loss of agrin strongly correlated with the expression of tenascin, an extracellular matrix molecule which has been described previously to be absent in mature non-pathological brain tissue and to accumulate in the basal lamina of tumor vessels. These results support the view that in human GBM, BBB breakdown is reflected by the changes of the molecular compositions of both the endothelial TJ and the basal lamina.
G. Rascher · A. Fischmann · H. Wolburg (✉) Institute of Pathology, University of Tübingen, Liebermeisterstrasse 8, 72076 Tübingen, Germany e-mail:
[email protected], http://www.med.uni-tuebingen.de/~hgwolbur F. Duffner · E.-H. Grote Department of Neurosurgery, University of Tübingen, Hoppe-Seyler-Strasse 3, 72076 Tübingen, Germany S. Kröger Institute for Physiological Chemistry and Pathobiochemistry, University of Mainz, Duesbergweg 6, 55099 Mainz, Germany
Keywords Human · Glioma · Blood-brain barrier · Tight junctions · Agrin
Introduction Glioblastoma multiforme (GBM) is the most malignant human brain tumor, with an average survival time of 6–12 months after primary diagnosis [27]. Common features responsible for GBM malignancy include rapid cell proliferation, invasion into surrounding brain tissue and a high angiogenic activity. Under healthy conditions, the blood-brain barrier (BBB) protects the brain from changing composition of the blood and warrants a stable microenvironment in the neural parenchyma. It is located in the endothelial cells and restricts the paracellular diffusion of hydrophilic molecules by complex tight junctions (TJ) and a low degree of transcytosis [26]. In GBM tumors, vessels lose their BBB characteristics [21, 22]. This leads to the most serious clinical sign of brain edema as a result of increased vessel permeability. The lack of BBB characteristics is most apparent in morphological alterations of GBM vessels such as fenestrations, the increase of caveolae, pericyte detachment and the thickening and alteration of the extracellular matrix (ECM) [3, 8, 11]. Many factors are known to be involved in the maintenance of the BBB, including the brain microenvironment [10, 34]. Astrocytic endfeet are in close proximity to capillaries and communicate across the basal lamina with endothelial cells to induce and maintain the BBB. The nature of communication between endothelial cells and astrocytes and the involvement of the basal lamina are poorly understood. The aim of the present study was, therefore, to evaluate whether changes of ECM components occur in GBM vessels and whether these changes may correlate with the loss or disturbance of the BBB. Seven cases of GBM and one control tissue were screened for the ECM molecules agrin and tenascin, and the BBB TJ molecules occludin, claudin-1 and claudin-5. Agrin, a heparan sulfate proteoglycan [29] was first isolated from the electric organ of Torpedo californica [24] and shown
86 Immunohistochemistry
to be essential for clustering acetylcholine receptors in the postsynaptic membrane of the motor endplate [21]. Recently, agrin was also found in the basement membranes of vessels with special barrier properties like those in the brain, thymus and testis [2]. Astrocytes have been shown to express agrin in vivo and in vitro, therefore agrin could serve as a factor to maintain barrier functions in vivo [2], and the secretion of agrin might be regulated by endothelial cells [20]. It is believed that in skeletal muscle agrin is linked to the cytoskeleton via dystroglycan and the dystrophin-glycoprotein complex [5]. Tenascin is a hexameric glycoprotein that plays an important role during development, tissue remodeling and disease. In adult vessels, tenascin is restricted to regions of neovascularization and wound healing. It is up-regulated in pathological conditions like hypertension and malignant conditions. It is believed to be important for cell adhesion, migration and proliferation and it correlates with angiogenic vessels in astrocytomas [35]. In the present study, we describe a mutually exclusive occurrence of tenascin and agrin immunoreactivity within the ECM of tumor blood vessels. Furthermore, a correlation of agrin deposition and occludin expression in vessels is demonstrated. These findings are consistent with the hypothesis that agrin is involved in the maintenance of the BBB and that angiogenic vessels expressing tenascin and losing agrin become permeable.
Cryostat sections (10–12 µm) were mounted on glass slides coated with poly-L-lysine, dried, and postfixed for 5 min in ethanol at 4°C, followed by acetone for 1 min at room temperature. For detection of antibody binding with fluorochrome-conjugated secondary antibody, the sections were rehydrated in TBS, blocked by incubation for 30 min with 5% (w/v) skimmed milk, 0.3% (w/v) Triton X-100 (Serva, Heidelberg, Germany) and 0.4% (w/v) NaN3 in TBS. Antibodies were diluted in the same solution and sections were incubated either overnight at 4°C or for 1 h at room temperature. After three washes in TBS for 10 min, sections were incubated for 45 min with the secondary antibody at room temperature. Following washes in TBS, sections were postfixed in buffered 4% paraformaldehyde, washed again in TBS and mounted in 90% glycerol/ 10% TBS. Fluorescence was visualized with a confocal laser scanning microscope (Axiovert 135M, Zeiss, Oberkochen, Germany) and images were processed using Adobe Photoshop (version 5.5, Adobe, Mountain View, Calif.). For detection of antibody binding with the immunoperoxidase method, the Vectastain universal ABC kit (Vector Laboratories, Burlingame, Calif.) with 3,3’-diaminobenzidine as chromogen was used according to the manufacturers instructions with the following modifications: quenching of endogenous peroxidase was performed by incubation with 10% methanol and 3% H2O2 in distilled water for 30 min and unspecific binding was minimized by incubation with the blocking-solution for 20 min as detailed above. The sections were counterstained with Mayer’s hematoxylin and eosin, dehydrated in a series of ethanol and mounted in Depex. Specimens were documented with an Axiophot photomicroscope (Axioplan, Zeiss) on Fuji T64 film material and images were processed as mentioned above.
Material and methods
Results
Small fragments of seven GBM specimens were isolated from patients at the Department of Neurosurgery of Tübingen University. In addition, as control tissue, a small fragment of cerebral cortex was isolated from a patient with traumatic injury. After surgically removal, the tissue was embedded unfixed in Tissue Tec O.C.T. compound and frozen in liquid nitrogen. The time between surgery and freezing ranged from 20 min (GBM) to 2 h (traumatic injury). All chemicals were purchased from Merck, Darmstadt, Germany, unless specifically mentioned.
Distribution of ECM compounds
Antibodies The production of the antibody against mouse claudin-5/TMVCF has been described previously [19]. Additionally, the following antibodies for immunohistochemistry were used: polyclonal rabbit anti-human fibronectin antiserum (Dako, Carpinteria, Calif.), FITC-conjugated polyclonal sheep anti-human fibronectin antiserum (Bio Trend Chemikalien, Köln, Germany), polyclonal rabbit anti-human claudin-1 antiserum and polyclonal rabbit anti-human occludin antiserum (Zymed Laboratories, San Francisco, Calif.), and polyclonal rabbit anti agrin antiserum (46–1; 17, 33). Monoclonal mouse anti-human tenascin antibody (Dako) was raised against tenascin-C; however, since some epitopes are also present in other tenascin isoforms, this antibody may recognize several if not all tenascins. However, Akhtar et al. [1] used this antibody for recognition of tenascin-C. Secondary antibodies labeled with cyanin-derivative dyes Cy2 and CY3 were purchased from Dianova (Hamburg, Germany), and secondary antibodies labeled with Alexa Fluor were purchased from Molecular Probes (Göttingen, Germany). For controls, the primary antibodies were omitted. In doublelabeling studies, controls included crossover incubation to exclude cross-reaction.
The vessels of seven GBM were screened for tenascin and agrin. Brain tissue from a traumatic injury served as a control. The distribution of tenascin and agrin was mutually exclusive: in the control tissue, no tenascin was found, whereas agrin could be detected in the basal lamina of all vessels. The histology of the small specimens showed the malignancy of the tumors as detailed in the Table 1 Summary of the immunohistochemical detection of the ECM molecules agrin, tenascin and the junction molecules occludin, claudin-5, claudin-1 in the control tissue and seven GMB. In GBM 18 and GBM 15 two different regions were analyzed (a, b). The strength of expression ranged from high (+++), medium (++), low (+) and no detection (–) (ECM extracellular matrix, GBM glioblastoma multiforme)
Control GBM 19 GBM 12 GBM 16 GBM 18a GBM 18b GBM 15a GBM 15b GBM 13 GBM 17
Tenascin Agrin
Occludin Claudin-5 Claudin-1
– – – – –/++ +++ – ++ +++ +++
+++ +++ +++ +++ +++ – +++ – +++/– –
+++ +++ +++ +++ +++ – +++ – +++/– –
+++/+ +/– +++ + + – +++ – –/+ –
–/+ –/+ –/+ –/+ – – –/+ – – –
87
Fig. 1 Immunohistochemical labeling of agrin (A, D, G) and tenascin (B, E, H) in serial native cryostat sections from the brain of a traumatic injury (A–C) and in two different GBM tissues (D–I). Staining control sections are shown in C, F, I. In the traumatic tissue (A–C), agrin labeling is found in vessels (arrow in A), but tenascin can not be detected (arrow in B). GBM 18 and 13 show tenascin labeling in vessels and in the surrounding tumor tissue (E, H). In GBM 18 both, agrin and tenascin labeling is found associated with vessels (arrows in D, E). However, agrin and tenascin labeling does not overlap (arrows in D, E). In GBM 13, tenascin labeling is primarily found in vessels (arrow in H), and only a faint agrin labeling is detected in the inner part of the vessel ECM (arrows in G) (GBM glioblastoma multiforme, ECM extracellular matrix)
neuropathological reports. GBM 19, 12 and 16 showed only agrin immunoreactivity and GBM 17 only tenascin immunoreactivity (Table 1). Tissue samples of GBM 18, 15 and 13 revealed that these tumors were highly heterogeneous with respect to their vasculature: vessel density and the amount of hyperplastic vessels were extremely high. In these tumors, region-specific patterns were detected, in which either agrin and no tenascin, or tenascin without agrin were observed. However, in some regions, both molecules were present (Fig. 1D, E) but did not colocalize within single vessels (Fig. 2C). Instead, tenascin
88 Fig. 2 Immunohistochemical labeling of GBM 19 (A, B), GBM 13 (C–E), and GBM 16 (F–H) in a series of sections with antibodies against tenascin, agrin, fibronectin, occludin, claudin-5 and claudin-1. A, B The vessels of GBM 12 express agrin but not tenascin. The occludin labeling in fibronectin-positive vessels is strictly junction-associated. C–E GBM 13 is highly heterogeneous. The left vessel shows tenascin labeling in the outer, agrin labeling in the inner part of the ECM, and a junction-associated occludin labeling (arrow in D). In contrast, the right vessel shows only a tenascin (C) and no occludin labeling (D). No claudin-5 could be detected in any vessels (E). F–H The vessels of the GBM 15 were very heterogeneous. In the region depicted, the vessels express agrin (F) and claudin-5 (G), but not claudin-1 (H). In G and H, the yellow spots do not represent vessel-related codistributions of claudin-5/1 and fibronectin, respectively
immunoreactivity could be detected at the outer – astroglial – tumor side and agrin at the inner – endothelial – side of the ECM (Fig. 2C, Table 1). All vessels that were labeled with an antibody against fibronectin in serial sections were never immunonegative for both agrin and tenascin. The extravascular labeling as produced by the
anti-tenascin antibody shown in Fig. 1E, H may be due to a specific expression of tenascin by tumor cells. However, we can not rule out an additional component of unspecific staining in the tumor tissue apart from the blood vessels. In the case of agrin, the faint staining of extravascular tumor tissue with the anti-agrin antibody (Fig. 1D, G) may
89 Table 2 Overview about the distribution of agrin, tenascin and tight junctional proteins as found in different glioma blood vessels Blood vessel type
Agrin
Occludin Claudin-5 Claudin-1 Tenascin
Type 1 Type 2 Type 3 Type 4
+ + + –
+ + + –
+ + – –
+ – – –
– –/+ +/– +
be due to an unspecific staining, because an expression of agrin by tumor cells has not been described in the literature so far. Expression of junctional proteins In a previous study, we investigated the alteration of TJ proteins in blood vessels of human GBM [19]. We found a down-regulation of claudin-1 in the endothelial TJs and a dislocation of claudin-5 from the junctional position to cytoplasmic vesicles. In the present study, we analyzed whether the loss of agrin is accompanied by an altered distribution of the TJ molecules occludin, claudin-1 and claudin-5. The junctional proteins were immunohistochemically detected and their expression was compared with that of agrin in the extracellular matrix of single vessels. As demonstrated in Fig. 2, a codistribution of occludin and agrin immunoreactivities was observed in serial sections (all four vessel types, Table2). In GBM 13, even directly adjacent vessels were found to express either agrin and occludin, or neither (Fig. 2C–E). However, in this tumor the expression of tenascin did not correlate with the disappearance of occludin (Fig. 2C–E, Table 1): vessels that were immunopositive for tenascin and agrin were also positive for occludin (vessel types 2 and 3, Table 2). Vessels that were immunopositive for tenascin and immunonegative for agrin did not express occludin (vessel type 4, Table 2). Vessels immunonegative for tenascin and immunopositive for agrin consistently expressed occludin (vessel type 1, Table 2). Furthermore, it was found that claudin-1 was lost from the majority of tumor microvessels. Anti-tenascin-immunoreactive blood vessels regularly lacked claudin-1 (vessel types 2–4, Table 2). All vessels expressing claudin-1 also expressed claudin-5 (Tables 1, 2vessel type 1). However, claudin-5 was only detected in vessels expressing occludin and agrin (vessel types 1, 2, Table 2), although not all vessels expressing occludin and agrin also expressed claudin-5 (Fig. 2F–H, Table 2 vessel type 3). Table 2 summarizes the distribution of the different antigens tested in GBM, classifying the vessels into four types.
Discussion In the present study, we demonstrate that, under the pathological condition of GBM, the ECM-component agrin is partially lost from the basal lamina of blood vessels. In
such vessels, tenascin is deposited in the basal lamina. In other vessel profiles, tenascin was found in the outer (glial) side and agrin in the inner (endothelial) side of the ECM. Since tenascin is normally not present in brain blood vessels [13], the segregated expression of both molecules in one vessel speaks in favor of the hypothesis that it reflects a state of a dynamic process during which the up-regulation (at least presence) of tenascin is connected to the down-regulation (at least absence) of agrin. Furthermore, we found a strict positive correlation of agrin deposition and the expression of the endothelial TJ molecule occludin, and a disappearance of claudin-5 and claudin-1 from the TJ even in some agrin-containing vessels (Table 2). The heparan sulfate proteoglycan agrin contains several sites which undergo alternative mRNA splicing, resulting in the generation of several agrin isoforms. Only specific agrin isoforms are responsible for clustering acetylcholine receptors in the postsynaptic membrane of the motor endplate of skeletal muscle [22, 28]. Recently, agrin transcripts have been detected in the central nervous system [31] and in the basal lamina of vessels with barrier properties, whereas peripheral vasculature does not contain agrin [2]. Astrocytes have been shown to express agrin in vivo and in vitro [30, 31], supporting the view that agrin could serve as a factor to maintain barrier functions in vivo, and its secretion may be regulated by endothelial cells [20]. Under pathological conditions such as Alzheimer’s disease (AD), the BBB becomes leaky. Here, agrin is associated with β-amyloid plaques in close proximity to blood vessels [7, 9], and the vascular basement membrane-associated agrin becomes fragmented. This raises the possibility that the agrin deposited in senile plaques originates from the vessel basal lamina. Furthermore, Berzin et al. [4] showed that microvascular changes were associated with the appearance of perivascular prothrombin immunoreactivity, which can only be detected in leaky vessels, indicating that agrin levels are altered in association with microvascular damage in AD. In glioma, the BBB is also disrupted [21, 23]. The present study is the first to describe the loss of agrin from the endothelial basal lamina in blood vessels of human gliomas. One of the best investigated properties of brain tumors is their angiogenic activity and its dependence on growth factors such as vascular endothelial growth factor (VEGF) or transforming growth factor-β (TGF-β) and basic fibroblast growth factor (bFGF). A correlation of tenascin expression, angiogenesis and high permeability of the BBB has been demonstrated [35]. In tumor cell lines, tenascin expression could be induced by a variety of polypeptide growth factors such as TGF β-1, VEGF, FGF-2 and tetradecanoyl phorbol acetate (TPA) (as reviewed by [13]), suggesting that the expression of tenascin in the vessels results from an induction by neighboring tumor cells. It is not clear which other cells express tenascin, but they might include pericytes, endothelial cells and astrocytes. Electron microscopical labeling and/or in situ hybridization experiments have to be performed to answer this question. During normal development tenascin expression
90
is restricted to the angiogenic period in which the BBB is still immature [13]. In brain tumors, histological malignancy has been correlated with the expression of tenascin [15, 25], and tenascin has been suggested to be an angiogenic modulator [12, 36]. Therefore, we speculate that the up-regulation of tenascin in tumor blood vessels may reflect or even induce a disruption of the BBB. As tenascin is up-regulated, agrin disappears from the ECM of GBM vessels. The question arises, which mechanism underlies the disappearance of agrin. During angiogenic processes, degradation of the perivascular ECM is mediated by soluble and membrane-type matrix-metalloproteinases (MMPs and MT-MMPs), produced by tumor and pericytic cells. Increased expression of different MMPs (MMP-9; MMP-2, MMP-12) has been shown in tumor cells [14], but few data exist about the expression of MMPs in vessels. One possibility for the removal of agrin is its cleavage by tumor-activated MMPs or by plasmin. MMP3 has been reported to cleave agrin in skeletal muscle [32], but it was shown not to be expressed in GBM [33]. An alternative explanation for the disappearance of agrin from GBM vessels is the possibility that the binding of tenascin to agrin [6] would block binding of the anti-agrin antibody. However, this possibility is unlikely, because the polyclonal antiserum used in this study was generated to the entire C-terminal half of the agrin molecule [17]. Regarding the expression patterns of occludin and agrin, we consistently observed a codistribution of both molecules in GBM vessels. If agrin was absent, occludin was not detectable (see Table 1). This supports the possibility that the expression of occludin is dependent on the presence of agrin in the ECM. The correlation between agrin and the claudins is less obvious. Claudin-1 is believed to represent the most important permeability-restricting molecule of TJs. Under different pathological or non-physiological conditions such as cell culture [18], experimental allergic encephalomyelitis (Hamm et al., submitted) and glioma ([19] and this study), claudin-1 is down-regulated in BBB endothelial cells. Unexpectedly, we were not able to detect claudin-1 even in the control tissue. The control tissue used in the present study came from a patient who suffered from a traumatic injury. We are aware of that this is not appropriate because traumatic lesions may affect BBB features. In addition, the material remained unfixed for about 2 hour between accident and fixation. During this interval, the structural integrity of the BBB could have been disturbed. We therefore consider the material at best a control tissue for the analysis of ECM compounds and not for the quality of the BBB TJ, because the ECM composition in this human material, at least regarding tenascin and agrin, is identical with that in normal rat brain [2, 13]. Considering claudin-5, there was a large variability in the expression levels in the tumors. Vessels containing agrin together with claudin-5 were observed as well as vessels with agrin lacking claudin-5. In any case, we detected no vessels with claudin-1 and without claudin-5, and no vessels with the claudins and without occludin. Therefore, it is tempting to speculate that during
the barrier impairment in the glioma claudin-1 is the most sensitive TJ protein which disappears from the junction. Claudin-5 would be the next to disappear, and occludin probably would be the last one to be lost from the junction. As agrin is absent from the vessel basal lamina, no TJ protein will be detected. This relationship is demonstrated in the Table 2. In conclusion, all findings of the present study seem to support the hypothesis that (1) tenascin is a component of a molecular system activating angiogenesis and migration and proliferation of cells during both development (reviewed by [13]) and neoplasia [35], and (2) agrin is a component of a molecular system activating differentiation and maturation of vascular barriers. Whereas tenascin is mainly expressed during early development, agrin is expressed in later stages of development, and concomitantly with BBB formation. The mechanisms by which agrin influences or even induces the formation of the BBB is completely unknown. It will be a challenge for future research to detect the role of agrin in the mechanisms of BBB regulation in health and disease. Acknowledgements This work was supported by a grant from the Federal Ministry of Education, Science, Research and Technology (Fö. 01KS9602) and the Interdisciplinary Center of Clinical Research (IZKF) Tübingen. The diagnostic work of the Institute of Brain Research of the University of Tübingen is gratefully acknowledged. We thank Dr. Holger Gerhardt for valuable comments on the manuscript, Dr. Hubert Kalbacher for the production of the claudin-5 antibody and Kristin Möckel for skillful technical assistance with the immunohistochemistry.
References 1. Akhtar S, Bron AJ, Hawksworth NR, Bonshek RE, Meek KM (2001) Ultrastructural morphology and expression of proteoglycans, ßig-h3, tenascin-C, fibrillin-1, and fibronectin in bullous keratopathy. Br J Ophthamol 85:720–731 2. Barber AJ, Lieth E (1997) Agrin accumulates in the brain microvascular basal lamina during development of the bloodbrain barrier. Dev Dyn 208:62–74 3. Bertossi M, Virgintino D, Maiorano E, Occhiogrosso M, Roncali L (1997) Ultrastructural and morphometric investigation of human brain capillaries in normal and peritumoral tissues. Ultrastruct Pathol 21:41–49 4. Berzin TM, Zipser BD, Rafii MS, Kuo-Leblanc V, Yancopoulos GD, Glass DJ, Fallon JR, Stopa EG (2000) Agrin and microvascular damage in Alzheimer’s disease. Neurobiol Aging 21:349–355 5. Blake DJ, Kröger S (2000) The neurobiology of Duchenne muscular dystrophy: lerning lessons from muscle? Trends Neurosci 23:92–99 6. Cotman SL, Halfter W, Cole GJ (1999) Identification of extracellular matrix ligands for the heparan sulfate proteoglycan agrin. Exp Cell Res 249:54–64 7. Cotman SL, Halfter W, Cole GJ (2000) Agrin binds to β-amyloid (Aβ), accelerates Aβ fibril formation, and is localized to Aβ deposits in Alzheimer’s disease brain. Mol Cell Neurosci 15:183–198 8. Dinda AK, Sarkar C, Roy S, Kharbanda K, Mathur M, Khosla AK, Banerji AK (1993) A transmission and scanning electron microscopic study of tumoral and peritumoral microblood vessels in human gliomas. J Neurooncol 16:149–158
91 9. Donahue JE, Berzin TM, Rafii MS, Glass DJ, Yancopoulos GD, Fallon JR, Stopa EG (1999) Agrin in Alzheimer’s disease: altered solubility and abnormal distribution within microvasculature and brain parenchyma. Proc Natl Acad Sci USA 96: 6468–6472 10. Hayashi Y, Nomura M, Yamagishi S, Harada S, Yamashita J, Yamamoto H (1997) Induction of various blood-brain barrier properties in non-neural endothelial cells by close apposition to co-cultured astrocytes. Glia 19:13–26 11. Hirano A, Matsui T (1975) Vascular structures in brain tumors. Hum Pathol 6:611–621 12. Jallo GI, Friedlander DR, Kelly PJ, Wisoff JH, Grumet M, Zagzag D (1997) Tenascin-C expression in the cyst wall and fluid of human brain tumors correlates with angiogensis. Neurosurgery 41:1052–1059 13. Jones FS, Jones PL (2000) The tenascin family of ECM glycoproteins: structure, function, and regulation during embryonic development and tissue remodeling. Dev Dyn 218:235–259 14. Kachra Z, Beaulieu E, Delbecchi L, Mousseau N, Berthelet F, Moumdjian R, Del Maestro R, Beliveau R (1999) Expression of matrix metalloproteinases and their inhibitors in human brain tumors. Clin Exp Metastasis 17:555–566 15. Kim CH, Bak KH, Kim YS, Ko Y, Oh SJ, Kim KM, Hong EK (2000) Expression of tenascin-C in astrocytic tumors: its relevance to proliferation and angiogenesis. Surg Neurol 54:235– 240 16. Kröger S (1997) Differential distribution of agrin isoforms in the developing and adult avian retina. Mol Cell Neurosi 10: 149–161 17. Kröger S, Horton SE, Honig LS (1996) The developing avian retina expresses agrin isoforms during synaptogenesis. J Neurobiol 29:165–182 18. Liebner S, Kniesel U, Kalbacher H, Wolburg H (2000) Correlation of tight junction morphology with the expression of tight junction proteins in blood-brain barrier endothelial cells. Eur J Cell Biol 79:707–717 19. Liebner S, Fischmann A, Rascher G, Duffner F, Grote E-H, Kalbacher H, Wolburg H (2000) Claudin-1 and claudin-5 expression and tight junction morphology are altered in blood vessels of human glioblastoma multiforme. Acta Neuropathol 100:323–331 20. Lieth E, Barber A, Gardner TW (1995) Astrocyte expression of brain microvascular agrin-related protein is induced by endothelial cells. Mol Biol Cell 6:337a 21. Long DM (1970) Capillary ultrastructure and the blood-brain barrier in human malignant brain tumors. J Neurosurg 32:127– 144 22. McMahan UJ (1990) The agrin hypothesis. Cold Spring Harb Symp Quant Biol 55:407–418
23. Nir I, Kohn S, Doron Y, Israel O, Front D (1986) Quantitative analysis of tight junctions and the uptake of 99mTc in human gliomas. Cancer Invest 4:519–524 24. Nitkin RM, Smith MA, Magill C, Fallon JR, Yao YM, Wallace BG, McMahan UJ (1987) Identification of agrin, a synaptic organizing protein from Torpedo electric organ. J Cell Biol 105: 2471–2478 25. Oz B, Karayel FA, Gazio NL, Ozlen F, Balci K (2000) The distribution of extracellular matrix proteins and CD44S expression in human astrocytomas. Pathol Oncol Res 6:118–124 26. Peters A, Palay SL, Webster H (1991) The fine structure of the nervous system, 3rd edn. Oxford University Press, New York 27. Puzzilli F, Ruggeri A, Mastronardi L, Di Stefano D, Lunardi P (1998) Long-term survival in cerebral glioblastoma. Case report and critical review of the literature. Tumori 84:69–74 28. Ruegg MA, Bixby JL (1998) Agrin orchestrates synaptic differentiation at the vertebrate neuromuscular junction. Trends Neurosci 21:22–27 29. Tsen G, Halfter W, Kröger S, Cole GJ (1995) Agrin is a heparansulfate proteoglycan. J Biol Chem 270:3392–3399 30. Tsen G, Napier A, Halfter W, Cole GJ (1995) Identification of a novel alternatively spliced agrin mRNA that is preferentally expressed in non-neural cells. J Biol Chem 270:15934–15937 31. Stone DM, Nikolics K (1995) Tissue- and age-specific expression patterns of alternatively spliced agrin mRNA transcripts in embryonic rat suggest novel developmental roles. J Neurosci 15:6767–6778 32. VanSaun M, Werle MJ (2000) Matrix metalloproteinase-3 removes agrin from synaptic basal lamina. J Neurobiol 43:140– 149 33. Vince GH, Wagner S, Pietsch T, Klein R, Goldbrunner RH, Roosen K, Tonn JC (1999) Heterogeneous regional expression patterns of matrix metalloproteinases in human malignant gliomas. Int J Dev Neurosci 17:437–445 34. Wolburg H, Neuhaus J, Kniesel U, Krauß B, Schmid E-M, Öcalan M, Farrell C, Risau W (1994) Modulation of tight junction structure in blood-brain barrier endothelial cells: effects of tissue culture, second messengers and cocultured astrocytes. J Cell Sci 107:1347–1357 35. Zagzag D, Friedlander DR, Miller DC, Dosik J, Cangiarella J, Kostianovsky M, Cohen H, Grumet M, Greco MA (1995) Tenascin expression in astrocytomas correlates with angiogenesis. Cancer Res 55:907–914 36. Zagzag D, Friedlander DR, Dosik J, Chikramane S, Chan W, Greco MA, Allen JC, Dorovini-Zis K, Grumet M (1996) Tenascin-C expression by angiogenic vessels in human astrocytomas and by human brain endothelial cells in vitro. Cancer Res 56:182–189