Decorin Gene Transfer-Mediated Suppression of TGF-␤ Synthesis Abrogates Experimental Malignant Glioma Growth in Vivo

Gene Therapy (1998) 5, 1187–1194  1998 Stockton Press All rights reserved 0969-7128/98 $12.00 http://www.stockton-press.co.uk/gt Decorin gene transfer-mediated suppression of TGF-␤ synthesis abrogates experimental malignant glioma growth in vivo

M Sta¨nder, U Naumann1, L Dumitrescu1, M Heneka1,PLo¨schmann1, E Gulbins2, J Dichgans1 and M Weller1 Department of 1Neurology and 2Institute of Physiology, University of Tuebingen, Germany

Cytokines such as transforming growth factor-␤ (TGF-␤) whereas all control animals die or have to be killed within are thought to mediate escape from immune surveillance 4 weeks. Decorin-expressing tumors show a four-fold in human malignant glioma. Here, we report that ectopic increase of inﬁltration by activated T cells and a 1.6-fold expression of the small TGF-␤-binding proteoglycan, deco- increase in total B and T cells. Chronic steroid-mediated ␤ ␤ rin, inhibits not only TGF- bioactivity but also TGF- 1 and immunosuppression abrogates the inhibitory effects of ␤ ␤ TGF- 2 mRNA transcription and TGF- protein synthesis decorin gene transfer. We conclude that decorin-induced by human LN-18, LN-229, T98G and rat C6 glioma cells in inhibition of TGF-␤ release by glioma cells signiﬁcantly vitro. Ectopic expression of decorin in C6 rat glioma cells enhances antiglioma immune responses in vivo. Clinical results in strong inhibition of tumor formation in vivo. Deco- evaluation of decorin gene therapy for human malignant rin-expressing C6 gliomas grow initially but regress to very gliomas may be warranted. small residual tumors at 12 weeks after implantation

Keywords: malignant glioma; transforming growth factor ␤; decorin; gene therapy; immune system; immune therapy

Introduction be a useful strategy to induce the regression of experimental gliomas in rodents. Human malignant gliomas are resistant to current therapy including surgery and irradiation, as well as adju- vant chemotherapy and immunotherapy. The outcome Results has not been significantly improved in recent years, and median survival is still in the region of 1 year. Since there Generation of glioma cell sublines engineered to is significant immunosuppression in human glioma express human decorin 1 patients, notably in the T cell compartment, glioma has Human decorin cDNA was cloned into the BCMGS neo become an attractive target for various types of immuno- expression plasmid11 and transfected into the glioma cells 2 therapy. Release of immunosuppressive factors such as by electroporation. Expression of the decorin transgene ␤ 3 TGF- which induce T cell apoptosis is one putative was assessed by Northern blot, PCR and immunoblot pathway of glioma-induced immunosuppression that has analysis. Strong decorin mRNA expression was readily been targeted successfully in experimental glioma mod- detected in all decorin-transfected cell lines by Northern 4 els by antisense strategies. More recently, human glioma blot analysis and PCR (Figure 1a, b). In contrast, control cells have been shown to express functional CD95 ligand transfectants exhibited a decorin cDNA fragment only at the cell surface which may also protect the tumor from after two linked PCR procedures in LN-229 cells but not 5,6 cytotoxic T cell attack. in the other cell lines (data not shown). Lysates from Decorin is a small chondroitin–dermatan sulfate decorin-transfected cells showed strong immunoreactiv- ␤ proteoglycan that binds various forms of active TGF- ity whereas neo control cells transfected contained either ␤ through its core protein and may abrogate TGF- bioac- very little (LN-229) or no (LN-18, T98G, C6) decorin 7,8 ␤ tivity. The inhibitory effects of decorin on TGF- detectable by immunoblot analysis (Figure 1c, upper activity have first been exploited therapeutically in panel). Decorin was released in significant amounts into 9,10 experimental glomerulonephritis. Based on the pivotal the supernatant of all decorin-transfected cells but was role in the pathogenesis of malignant glioma attributed not detected in the supernatant of control transfectants ␤ to TGF- , we asked whether decorin gene transfer may (Figure 1c, lower panel). C6 decorin clone C4 was among the clones with the strongest decorin mRNA and protein expression, as well as decorin release into the supernatant. Proliferation, as assessed by determining gener- Correspondence: M Weller, Laboratory of Molecular Neuro-Oncology, Department of Neurology, University of Tuebingen, Medical School, ation times during logarithmic growth, in decorin-trans- Hoppe-Seyler-Strasse 3, 72076 Tuebingen, Germany fected pooled or cloned glioma cells did not differ from Received 17 November 1997; accepted 17 March 1998 that of pooled control transfectants in either cell line Decorin gene therapy for glioma M Sta¨nder et al 1188

Figure 1 Ectopic decorin expression in glioma cells. Total RNA or soluble protein lysates or supernatant proteins were obtained from control transfectants or pooled decorin-transfected cells of LN-18, LN-229, T98G and C6 glioma cell lines and subjected to PCR (a), Northern blot (b) or immunoblot analysis (c) as described in Materials and methods. For C6 cells, Figure 2 Growth and response to serum deprivation of decorin-trans- decorin clone C4, which was used for further studies, is also shown in a fected cells in vitro. (a) Control (neo) transfectants or decorin-transfected and c. In a, ␤-actin (lower panel) served as control, ‘n’ is a negative cells of LN-18, LN-229, T98G or C6 glioma cell lines, or C6 decorin clone control lacking RT in the RT reaction, and ‘p’ shows ampliﬁcation from C4, were seeded at 104 cells per well in 12-well plates in DMEM contain- the plasmid as positive control. In b, ␤-actin mRNA (lower panel) was ing 10% FCS. The number of living cells was determined at 24, 48, 72, detected to control for RNA integrity and equal loading. Note that deco- 96, 120 or 144 h after seeding using trypan blue staining and a hemocyto- rin-transfected cells showed not only the predicted mRNA species of 1.7 meter. Data are means (n = 8, s.e.m. Ͻ10%). (b) The cells were grown in kb, but a second larger mRNA species migrating at around 4 kb, presum- DMEM containing 10% FCS for 24 h, switched to serum-free medium ably derived from aberrant transcription from the plasmid. Consistent for 24, 48, 72, 96 or 120 h, and allowed to recover for 120, 96, 72, 48 or with the PCR data, no endogenous decorin mRNA was detected in neo 24 h, respectively, and stained by crystal violet assay. Cells grown in control cells. In c, the upper panel shows whole cell lysates and the lower FCS-containing medium continuously for 144 h served as 100% reference panel supernatant proteins. value. Data are expressed as mean percentages of growth/survival relative to this value (n = 4, s.e.m. Ͻ10%).

(Figure 2, left panel). Further, decorin-transfected cells did not differ from control transfectants in their resist- from human and rat decorin-transfected glioma cells ance to serum deprivation (Figure 2, right panel). inhibited the growth of TGF-␤-sensitive CCL64 cells significantly less efficiently than supernatants from parental or Effects of ectopic decorin expression on TGF-␤ neo control cells (data not shown). Specific ELISA assays ␤ ␤ bioactivity and RNA and protein synthesis were then performed to measure TGF- 1 and TGF- 2 ␤ Next we asked whether ectopic expression of decorin in the release into the supernatant (Figure 3a). TGF- 2 is the domi- glioma cells resulted in altered TGF-␤ activity. Supernatants nant active form of TGF-␤ released by the glioma cells. The Decorin gene therapy for glioma M Sta¨nder et al 1189

␤ ␤ Figure 3 Ectopic expression of decorin inhibits TGF- synthesis by malignant glioma cells. (a) Active (left panel) or total (right panel) TGF- 1 or TGF- ␤ 2 protein in the cell culture supernatant of decorin-transfected cells (closed bars) or control cells (open bars) was measured by ELISA. Data are expressed as means and s.e.m. of the TGF-␤ amount released by 106 cells in 24 h (*P Ͻ 0.05, **P Ͻ 0.01, t-test). (b) Soluble proteins of whole cell lysates (upper ␤ ␤ ␤ panel) or soluble supernatant proteins (lower panel) were analyzed for immunoreactive TGF- by immunoblot analysis. (c). TGF- 1 or TGF- 2 mRNA expression was assessed by RT-PCR. Data were normalized to the density of a coampliﬁed ␤-actin cDNA fragment and are expressed as ratios of optical ␤ + ␤ densities of TGF- 1 2 versus -actin PCR fragments.

␤ highest total level of TGF- 1 was found in LN-229 cells. The in C6 and T98G, moderate in LN-18 and hardly detectable ␤ ␤ highest total level of TGF- 2 was detected in LN-18 and in LN-229 cells. However, TGF- levels were dramatically T98G cells. Decorin gene transfer induced a strong reduced in the supernatants of the decorin-transfected cells ␤ ␤ ␤ reduction of active TGF- 2 and total TGF- 1 and TGF- 2 in compared with neo controls. We assume that this loss of ␤ rat and human glioma cells. The levels of TGF- 2 were immunoreactivity is not a result of complex formation of reduced at least 70% (active) and 55% (total). C6 decorin decorin and TGF-␤, preventing TGF-␤ from being detected clone C4 showed a reduction exceeding 90%. Immunoblot on immunoblot analysis. This is because SDS-PAGE is pre- analysis using reducing SDS gels confirmed that loss of dicted to disrupt noncovalent protein–protein interactions, TGF-␤ protein was not due to TGF-␤/decorin complex for- as exmplified here by the detection of the 12.5 kDa mono- mation, preventing antibody capture, but truely to a mer of TGF-␤ rather than the 25 kDa. reduction of TGF-␤ protein (Figure 3b). The analysis of That decorin not simply represses TGF-␤ bioactivity by cellular lysates which represent the intracellular pool of complex formation but by altering TGF-␤ mRNA tran- TGF-␤ revealed a reduction of the 12.5 kDa TGF-␤ mono- scription or stability, was confirmed by RT-PCR mer in decorin-transfected cells that was most prominent (Figure 3c) and Northern blot analysis (data not shown). Decorin gene therapy for glioma M Sta¨nder et al 1190

Figure 4 Ectopic expression of decorin abrogates glioma growth in vivo. 10 × 103 (a) or 2.5 × 103 (b–d) decorin-transfected (a, pooled decorin transfectants; b-d, clone C4) or neo C6 glioma cells were injected intrastriatally and the tumor volume determined on hematoxylin/eosin-stained sections using the MCID digitalization system at 2 (a, b) or 4 (c) weeks. (d) The kinetics of tumor growth over time is summarized. Data are expressed as means of tumor volumes and s.e.m. (**P Ͻ 0.01, t-test; n = 3/a, n = 4/b, n = 6/c).

Effects of ectopic decorin expression on tumor growth in 2.5 × 103 logarithmic-phase decorin-transfected or control vivo C6 cells intrastriatally and treated the animals with the To assess the effects of forced decorin expression on gli- steroid, dexamethasone, which is commonly adminis- oma cell growth in vivo, we chose the C6 rat glioma tered to human glioma patients for edema control and model. Several subclones of decorin-transfected C6 cells induces immunosuppression in these patients. Interest- were generated by limiting dilution cloning and charac- ingly, at a dose that is clinically used to induce T cell terized for decorin transgene expression by immunoblot suppression during transplantation (1 mg/kg), dexame- analysis. Pooled neo or decorin transfectants and decorin thasone completely reversed decorin-mediated tumor clone C4, which was among the best expressing clones, regression (Figure 6a and b). This was not due to a loss were used for the in vivo experiments. In the first experi- of decorin transgene expression (Figure 6c). We also ment, we injected 10 × 103 logarithmic-phase decorin- examined whether the in vivo effects of dexamethasone transfected or control C6 cells into the right striatum of might be the result of direct modulation of tumor cell 6-week-old female Sprague–Dawley rats and determined growth by dexamethasone and whether such actions of tumor volumes 2 weeks later. Figure 4a shows that deco- dexamethasone were modulated by ectopic decorin rin-transfected C6 gliomas had grown to much smaller expression. Growth curves and serum response challenge sizes than control cells. Next, 2.5 × 103 cells were injected experiments corresponding to the data in Figure 2 and the animals killed at 2, 4 or 12 weeks. At 2 weeks, showed that dexamethasone at 1 ␮m had no effect on the volume of C6 decorin gliomas was again significantly growth (Figure 6d) and resistance to serum deprivation reduced compared with C6 neo control animals (data not shown) either in decorin-transfected or neo (Figure 4b). At 4 weeks, there was regression to small control cells. residual tumors in the decorin-transfected animals. In contrast, the control animals had developed large tumors Discussion of around 70 mm3 (Figures 4c and 5a–c). All control animals had to be killed for progressive tumor growth at 4 The present study shows that ectopic expression of the weeks. One group of animals injected with decorin-trans- small proteoglycan, decorin, in malignant glioma cells fected tumor cells was observed for 12 weeks. Here, we inhibits TGF-␤ synthesis in vitro and promotes experi- found a small residual tumor of approximately the same mental glioma regression in vivo. Abrogation of glioma- volume as after 4 weeks (Figure 4d). We are also per- induced, TGF-␤-mediated immunosuppression in the forming a long-term study observing six animals injected host is most likely the therapeutic principle of decorin in early June 1997. These animals are still free of clinical gene transfer for malignant glioma since decorin-express- evidence for tumor relapse at the time of the final sub- ing tumors grew initially but later regressed and showed mission of this paper (February 1998). enhanced infiltration by immune cells and since the effect Persistent detection of transgene expression in vivo was of decorin was nullified by coexposure to the immuno- confirmed by immunochemistry. Decorin was deposited suppressive steroid, dexamethasone. Although TGF-␤ mainly in the extracellular matrix. No decorin was antisense strategies have also been used to limit glioma detected in C6 control gliomas (Figure 5d and e). Next, growth in experimental studies,4 decorin gene transfer we asked whether regression of decorin tumors was asso- may be preferable because of its broad activity against ciated with different patterns of host cell infiltration. various types of TGF-␤ which are expressed by glioma Immunochemical studies using marker antibodies to cells.12,13 CD45 (B and T cells), CD4 (T helper cells), CD8 (cytotoxic Proteoglycans are a class of extracellular matrix macro- T cells) and CD25 (activated T cells) were performed molecules composed of a core protein covalently linked (Figure 5f–j). There was a greater number of infiltrating to at least one glycosaminoglycan chain. Decorin consists cells expressing each marker in decorin C6 tumors. of a core protein of 45 kDa and one chondroitin/ derma- Notably the number of activated T cells was increased. tan sulfate chain and is expressed in human and rat To obtain further evidence for an immune-mediated embryonic and adult brain, eg in endothelial cells, vascu- regression of decorin-expressing gliomas, we injected lar smooth muscle cells, the tela choroidea, the ependyma Decorin gene therapy for glioma M Sta¨nder et al 1191 a d

b e

c f

Figure 5 Decorin gene transfer for glioma: tumor regression, transgene expression and host response. 2.5 × 103 pooled decorin (C4, dec) or neo (neo) C6 glioma cells were injected intrastriatally and the tumor volume and the immune response determined at 4 weeks. (a–c) hematoxylin/eosin-stained coronar sections of C6 neo tumor (a, ×32), C6 decorin tumor (b, ×32; c, ×100). (d and e). Decorin expression of C6 neo (d) and C6 decorin (e) tumors detected by immunohistochemistry (×1000). (f) CD45-positive cells in C6 decorin glioma (×1000). (g–j) CD45-, CD4-, CD8- or CD25-positive cells were counted in four different coronar sections each of 4 week decorin or control tumors (n = 3). Data were normalized to control tumors and are expressed as means and s.e.m. (**P Ͻ 0.01, t-test). and in association with myelinated ﬁbres, and also in the The core decorin protein shows high binding afﬁnity cytoplasm of neurons and astrocytes.14–16 Decorin for TGF-␤ and has been hypothesized to act as a reservoir expression is enriched in boundary regions around blood for TGF-␤.17 Decorin may inhibit TGF-␤ activity by com- vessels, brain surface, ependyma and plexus choroideus, peting with TGF-␤ receptors for a common or adjacent and is enhanced in response to brain injury.14 binding site.8 More recently, decorin has been shown to Decorin gene therapy for glioma M Sta¨nder et al 1192

Figure 5 Continued.

be part of a negative feed back loop that regulates TGF- Protein detection ␤ activity. Decorin may directly suppress the expression Immunoblot analysis was performed as described pre- of TGF-␤ mRNA,10 whereas TGF-␤ stimulates the viously.19 Briefly, cells were lysed in lysis buffer (50 mm expression of decorin mRNA and protein in several cell Tris-HCl, pH 8, containing 120 mm NaCl, 0.5% NP-40, 2 lines and tissues.17 We find that decorin represses release ␮g/ml aprotinin, 100 ␮g/ml PMSF, 10 ␮g/ml leupeptin, and synthesis of TGF-␤ protein and decreases TGF-␤ 50 mm sodium fluoride) followed by high-speed centri- mRNA in human and rat glioma cells (Figure 3a–c). Gli- fugation at 4°C for 10 min. The soluble lysates were used oma cell growth in vitro and tolerance of serum depri- for TGF-␤ immunoblot analysis. For decorin labeling, an vation were unaffected by ectopic expression of decorin, aliquot of 20 ␮g protein was precipitated with acetone, suggesting that TGF-␤ is not a survival factor or modu- dried and collected in digestion buffer (25 mm Tris-HCl, lator of growth for these cells in vitro. These findings pH 8, 30 mm sodium acetate, 10 mm EDTA, 10 mm N- contrast with several other neoplastic cell lines which ethylmaleimide, 5 mm PMSF, 0.36 mm pepstatin). Cell were growth-inhibited by decorin in a p21-dependent culture supernatants were treated accordingly. The manner.18 samples were digested with chondroitinase ABC (0.1 Decorin-expressing C6 cells formed a visible tumor U),22 separated by 12% SDS-PAGE, and electroblotted to that regressed over time whereas control cells formed nitrocellulose. Immunodetection involved blocking for 1 large tumors that killed their host animals within 4 weeks h in blocking solution (10 mm Tris-HCl, pH 7.5, contain- (Figure 4b and c). No tumor recurrence has been noted ing 150 mm NaCl, 0.1% Tween 20, 5% skim milk, 2% clinically in six animals followed for 8 months now. BSA), incubation with decorin or TGF-␤ antibody (1 In an effort to elucidate the mechanism of decorin- ␮g/ml), overnight at 4°C, and labeling with alkaline induced glioma regression, we determined that the num- phosphatase-conjugated anti-rabbit IgG (1:10 000 in ber of infiltrating B and T cells was significantly higher PBS/0.1% Tween 20) and nitroblue tetrazolium chloride in decorin-expressing tumors than in control tumors. (0.41 mm) and 5-bromo-4-chloro-3-indolyl phosphate Further, the synthetic steroid, dexamethasone, which did (0.38 mm) in 200 mm Tris-HCl, pH 9.5, containing 10 mm not differentially modulate the growth of control or deco- MgCl2 as substrate. rin-transfected C6 cells in vitro, nullified the effects of decorin on tumor growth in vivo. These observations raise RNA analysis the possibility that decorin-mediated inhibition of TGF- Decorin and TGF-␤ mRNA expression were studied by ␤ synthesis enhances the immunogenicity of malignant Northern blot analysis and PCR. For Northern blot analy- glioma cells in vivo and that decorin gene transfer holds sis, RNA was denatured, separated by electrophoresis on promise as a new approach of somatic immune gene 1.5% agarose gels (20 ␮g per lane), and transferred to therapy for these lethal neoplasms. nylon membranes. Nucleic acids were immobilized by ultraviolet irradiation. Membranes were prehybridized Materials and methods with 50% formamide, 1% SDS and 1 × SSC, and hybridized with DNA probes generated with 32P-dCTP Cell culture and generation of glioma cell sublines by random oligonucleotide priming. The blots were engineered to express decorin washed in 1 × SSC/1% SDS at 55°C for 15 min. Autoradi- LN-18 and LN-229 human malignant glioma cells were ography films were scanned by laser densitometry. Sig- kindly provided by Dr N de Tribolet (Lausanne, nals obtained with an actin cDNA probe served as the Switzerland). Human T98G and rat C6 glioma cells were control for equal loading. from ATCC (Rockville, MD, USA). The cells were grown For PCR analysis, mRNA was translated into cDNA in DMEM containing 10% FCS, 1 mm glutamine and 1% using the Qiagen reverse transcriptase kit (Qiagen, penicillin/streptomycin (‘culture medium’).19 Human Hilden, Germany). The PCR conditions were as follows: 11 ␤ ° ° ° decorin cDNA was cloned into BCMGS-neo. BCMGS- TGF- 1 35 cycles, 45 s/94 C, 45 s/55 C, 60 s/72 C, primer decorin or BCMGS-neo without insert, as a control, were sequence ACTGGTGCT GACGCCTGGC (nucleotides transfected into LN-18, LN-229, T98G or C6 cells by 895–914) and CCTTGCTGT ACTGCGTGTCC 20,21 ␤ ° electroporation. (nucleotides 1841–1860); TGF- 2 35 cycles, 45 s/94 C, 45 Decorin gene therapy for glioma M Sta¨nder et al 1193 s/55°C, 60 s/72°C, primer sequence ATGCAC- a TACTGTGTGCTGAGC (nucleotides 181–201) and CTGATCACCACTGGTATATGTG (nucleotides 954–975); decorin 35 cycles, 45 s/94°C, 45 s/52°C, 60 s/72°C, primer sequence CCCAGAAGTTCCTGATCAC (nucleotides 19–38) and CAGAGCGCACGTAGACAC (nucleotides 946–964). The resulting PCR fragment sizes are 966 ␤ ␤ bp (TGF- 1), 795 bp (TGF- 2) and 946 bp (decorin). PCR fragments were separated in 1% agarose gels and vis- ualized by ethidium bromide. A ␤-actin cDNA fragment was amplified as an internal control for equal amplifi- cation. For densitometric calculations, the ethidium bromide fluorescences were digitalized. For detection of endogen decorin mRNA of control cells, we used two PCR procedures of 25 cycles with the same temperatures and primers as above. b TGF-␤ assays TGF-␤ bioactivity was measured by bioassay using the TGF-␤-sensitive CCL64 mink lung epithelial cell line.23 Briefly, 5 × 103 cells were adhered to 96-well plates for 24 h, washed and exposed to diluted cell culture supernatants for 36 h. 3H-thymidine incorporation (1 ␮Ci per well) was assessed from 24 to 36 h. Recombinant TGF-␤ served as an internal standard. To confirm the specificity of growth inhibition by TGF-␤, appropriate neutraliz- ation studies were performed with neutralizing pan-TGF- ␤ antibody (25 ␮g/ml) (R&D, Minneapolis, MN, USA). Latent TGF-␤ was activated by transient acidification.13,24 ␤ ␤ TGF- 1 and TGF- 2 were also measured by specific ELISA (R&D) using the given protocol. For total TGF-␤, supernatants were transiently acidified and diluted 1:4 in assay buffer.

In vivo studies and immunochemistry Sprague–Dawley rats (Charles River, Sulzfeld, Germany) were housed in groups of three or four under standard conditions at a temperature of 22°C and a 12 h light-dark c cycle. They had free access to standard food pellets and tap water. For surgery, 12-week-old female rats weighing 250–300 g were anesthetized with pentobarbital and logarithmic-phase decorin-transfected or control C6 cells were injected stereotactically into the striatum. At 2, 4 or 12 weeks, the rats were killed. The brains were ﬁxed in formalin/acetic-acid/methanol (1/1/8), embedded in parafﬁn and cut in 8-␮m sections for histology or 20-␮m sections for volumetry. Tumor volumes were determined on hematoxylin/eosin-stained sections with the MCID digitalization system (Imaging Research, Ontario, Canada). The following monoclonal primary antibodies were used for immunohistochemistry: anti-CD4 (1:200), anti-CD8 (1:500), anti-CD25 (1:500) and anti-CD45 (1:300) d (Cedar Lane, Ontario, Canada). For decorin staining, the

Figure 6 Decorin-mediated abrogation of C6 glioma growth in vivo is abrogated by dexamethasone. (a) 2.5 × 103 pooled decorin or neo C6 glioma cells were injected intrastriatally and the rats treated daily with dexamethasone at 1 mg/kg. Tumor volume was determined at 4 weeks. Data are mean and s.e.m. (n = 3). (b) Large C6 decorin tumor at 4 weeks in a dexamethasone-treated animal (×32, compare with Figure 5b). (c) Decorin expression is unaffected by dexamethasone at 4 weeks (×400, compare with Figure 5e). (d) The growth of neo control cells (open symbols, dashed lines) or C6 clone C4 (filled symbols, straight lines) was monitored in the absence (circles) or presence (squares) of dexamethasone (1 ␮M)asin Figure 2a. Decorin gene therapy for glioma M Sta¨nder et al 1194 brains were frozen immediately in isopentane and stored 8 Hildebrand A et al. Interaction of the small interstitial proteogly- at −80°C. Tissue sections of 8 ␮m were cut with a Leica cans biglycan, decorin and fibromodulin with transforming ␤ CM3000 cryomicrotome (Leica, Nussloch, Germany), growth factor . Biochem J 1994; 302: 527–534. mounted on poly-l-lysine-coated glass slides and air- 9 Border WA et al. Natural inhibitor of transforming growth factor-␤ protects against scarring in experimental kidney disease. dried. The fixed tissue sections were pretreated with Nature 1992; 360: 361–364. chondroitinase ABC (Boehringer, Mannheim, Germany) 10 Isaka Y et al. Gene therapy by skeletal muscle expression of and stained with antidecorin antibody (1:400) (Novus decorin prevents fibrotic disease in rat kidney. Nature Med 1996; Molecular, San Diego, CA, USA). Bound primary anti- 4: 418–423. bodies were labeled with biotinylated antimouse (CD4, 11 Karasuyama H, Kudo A, Melchers F. The protein encoded by CD8, CD25, CD45) or antirabbit (decorin) IgG antibodies the VpreB and ␭5 pre-B cell genes can associate with each other (1:200; Camon, Wiesbaden, Germany), washed, incubated and with ␮ heavy chain. J Exp Med 1990; 172: 969–972. with avidin-biotin reagent (Vectastain Elite ABC peroxi- 12 Olofsson A et al. Transforming growth factor-␤1, ␤2 and ␤3 dase; Vector, Wiesbaden, Germany) and developed in secreted by human glioblastoma cell lines. J Biol Chem 1992; 267: diaminobenzidine (Sigma). For all antibodies, controls 19481–19488. 13 Constam B et al. Differential expression of transforming growth were performed by treatment with whole preimmune ␤ ␤ ␤ factor- 1, 2 and 3 by glioblastoma cells, astrocytes and serum. microglia. J Immunol 1992; 148: 1404–1410. 14 Stichel C et al. Differential expression of the small Acknowledgements chondroitin/dermatan sulfate proteoglycans decorin and biglycan after injury of the adult rat brain. Brain Res 1995; 704: This study was supported by the Else-U¨ belmesser-Foun- 263–274. dation and the Deutsche Krebshilfe. The authors thank E 15 Hanemann C et al. Expression of decorin mRNA in the nervous Ruoslahti (San Diego, CA, USA) for the human decorin system of rat. J Histochem Cytochem 1993; 41: 1383–1391. cDNA and Dr U Malipiero and Dr A Fontana for 16 Voss B, Glo¨ssl J, Cully Z, Kresse H. Immunocytochemical inves- helpful discussions. tigation on the distribution of small chondroitin sulfate-dermatan sulfate proteoglycans in human. J Histochem Cytochem 1986; 34: 1013–1019. References 17 Ruoslathi E, Yamaguchi Y. Proteoglycans as modulator of 1 Roszman T, Elliott L, Brooks W. Modulation of T cell function growth factor activities. Cell 1991; 64: 867–869. by gliomas. Immunol Today 1991; 12: 370–374. 18 Santra M et al. Ectopic expression of decorin protein core causes 2 Weller M, Fontana A. The failure of current immunotherapy for a generalized growth suppression in neoplastic cells of various malignant glioma. Tumor-derived TGF-␤, T cell apoptosis, and histogenetic orgin and requires endogenous p21, an inhibitor of the immune privilege of the brain. Brain Res Rev 1995; 21: cyclin-dependent kinases. J Clin Invest 1997; 100: 149–157. 128–151. 19 Weller M et al. Anti-Fas/APO-1 antibody-mediated apoptosis of 3 Weller M, Constam D, Malipiero U, Fontana A. Transforming cultured human malignant glioma cells. Induction and modu- ␤ growth factor- 2 induces apoptosis of murine T cell clones with- lation of sensitivity by cytokines. J Clin Invest 1994; 94: 954–964. out down-regulating bcl-2 mRNA expression. Eur J Immunol 20 Weller M et al. Proto-oncogene bcl-2 gene transfer abrogates 1994; 24: 1293–1298. Fas/APO-1 antibody-mediated apoptosis of human malignant 4 Fakhrai H et al. Eradication of established intracranial rat gli- glioma cells and confers resistance to chemotherapeutic drugs omas by transforming growth factor ␤ antisense gene therapy. and therapeutic irradiation. J Clin Invest 1995; 95: 2633–2643. Proc Natl Acad Sci USA 1996; 93: 2909–2914. 21 Weller M et al. Fas/APO-1 gene transfer for human malignant 5 Saas P et al. Fas ligand expression by astrocytoma in vivo: main- glioma. Cancer Res 1995; 55: 2936–2944. taining immune privilege in the brain? J Clin Invest 1997; 99: 22 Oike Y et al. Stuctural analysis of chick-embryo cartilage proteo- 1173–1178. glycan by selective degradation with chondroitin lyases 6 Weller M et al. CD95-dependent T cell killing by glioma cells (chondroitinases) and endo-␤-d-galactosidase (keratanase). expressing CD95 ligand: more on tumor immune escape, the Biochem J 1980; 191: 193–207. CD95 counterattack, and the immune privilege of the brain. Cell 23 Garrigue-Antar L et al. Optimisation of CCL64-based bioassay Physiol Biochem 1997; 7: 282–288. for TGF-beta. J Immunol Meth 1995; 186: 267–274. 7 Yamaguchi Y, Mann D, Ruoslathi E. Negative regulation of 24 Esser P et al. Apoptosis in proliferative vitreoretinal disorders. transforming growth factor-␤ by the proteoglycan decorin. Possible involvement of TGF-␤-induced RPE cell apoptosis. Exp Nature 1990; 346: 281–284. Eye Res 1997; 65: 365–378.