Journal o f Neuro-Oncology 18: 9-17,1994. 9 1994 Kluwer Academic Publishers. Printed in the Netherlands.

Laboratory Investigation

Interferon effect on glycosaminoglycans in mouse glioma in vitro

Marzenna Wiranowska and Abhinender K. Naidu

University of South Florida, College of Medicine, Department of Neurology, Tampa, Florida, USA

Key words: glioma, mouse interferon o~/13(MuIFN cz/[3), glycosaminoglycans (GAGs), chondroitin sulfate (CS), proteoglycans (PGs)

Summary The effect of mouse interferon c~/[3(MuIFN c~/[~)on the production of glycosaminoglycans (GAGs) by mouse glioma G-26 in vitro was evaluated. Two G A G species secreted extracellularly by the mouse glioma G-26 were isolated using cellulose acetate electrophoresis. They were identified as hyaluronic acid (HA) and chondroitin sulfate (CS) following enzymatic digestion with enzymes: hyaluronidase and chondroitinase ABC. Further characterization of CS by enzymatic digestion with specific chondroitinases for chondroitin 4-sulfate (CSA) and chondroitin 6-sulfate (CSC), revealed that the isolated CS was neither CSA nor CSC. Therefore, it may be either chondroitin sulfate B (CSB) (dermatan sulfate) or one of the 'chontroitin sulfate isomers' (D-H). The three day incubation of glioma G-26 cells with 8 x 10-8 x 104 U/ml of MuIFN (z/J3resulted in a dose dependent inhibition of cell proliferation measured by 3H-thymidine incorporation and the MTT assay. The significant decrease of the CS (p < 0.008) but not the H A level, (measured densitometrically), was observed following 72 hours (hrs) incubation of G-26 cells with 8 x 103 U/ml of MuIFN a/13 (IFN treated cells: 0.03 + 0.007 integrated optical density (IOD); control cells: 0.07 + 0.01 IOD). The decreased CS production may be the underlying cause of IFN mediated inhibition of glioma cell proliferation.

Introduction Therapy of malignant tumors of the central nervous system (CNS), e.g., glioma, is limited due to the blood-brain barrier (BBB) and blood-tumor barrier [1, 2]. In addition, it also has been shown that glioma cells similar to various carcinoma and sarcoma cells release an extracellular glycosaminoglycan (GAG) 'coat' which may constitute a barrier for a host immune response [3, 4], interfering with brain tumor therapy. Glycosaminoglycans are produced by a variety of cells [5, 6] including CNS neurons and glial cells [7]. Glycosaminoglycans are the only matrix components which are extracellularly synthesized by the CNS. They are localized on cell membranes and in the extracellular space [7, 8]. The glycosaminogly-

can family includes several major classes of molecules found in the CNS, such as nonsulfated hyaluronan, also known as hyaluronic acid (HA), and sulfated GAGs including: chondroitin 4-sulfate (CSA), chonodroitin 6-sulfate (CSC), chondroitin sulfate B (CSB), also known as dermatan sulfate, chondroitin sulfate isomers (D-H), keratan sulfate and heparan sulfate (HS) [9]. They play an essential role in cellular attachment, provide cellular support and regulate biosynthesis, as well as cell proliferation and differentiation [10]. It has been shown that in CNS tumors the production of GAGs is elevated [8] and these are histochemically different than GAGs produced by normal cells of the CNS [7,11]. In glioma the G A G level and the predominance of particular G A G species may depend on the type of glioma or the degree of anaplasia [7, 8, 11]. The

10 blood vessels of all CNS tumors were shown to contain GAGs [8, 11]. Their presence in vessels may contribute to the barrier between the blood and the glioma [11]. The elevated levels of extracellularly secreted G A G in glioma may both create an additional barrier for therapy and alter the extracellular basement membrane, facilitating further glioma invasion. Therefore, the detection of high G A G levels in the areas of the cortex infiltrated by the tumor [11] supports this concept. The elevated GAGs may also be covering glioma surface antigens preventing host mediated cellular immune responses [4]. The interferons (IFNs), which have a broad spectrum of biological activities, (anticellular, antiproliferative, antitumor, including the antiglioma and immunomodulatory) were shown to affect synthesis of cell membrane components and extracellular matrix macromolecules [12, 13]. For example, IFN has been shown to affect fibronectin distribution [12]; G A G induction in normal human fibroblasts [10,13], endothelial cells [14] and rheumatoid arthritis derived synovial cells [15] as well as collagen synthesis in both normal and scleroderma fibroblasts [13]. Thus far, the only known study of IFN effect on CNS cell membrane components showed that IFN caused altered production of fibronectin in normal rat brain and in rat glioma cells [16]. This study evaluates whether mouse IFN a/J3 (MuIFN ~/[~) which inhibits glioma growth in vitro, has an effect on synthesis of G A G by G-26 cells.

Materials and methods

Ce//s The in vitro continuously grown mouse glioma-26 (G-26) cell line, established in this laboratory, was used in this study. The G-26 cell culture was developed using G-26 tumor fragments from subcutaneously (sc) maintained glioma in C57BL/6 mice as previously described [17].

M u l F N off[3effect on G-26 proliferation and viability in vitro Established G-26 cell cultures were trypsinized, cells counted using trypan blue exclusion assay and placed in 96 well microtiter plates (Costar, Cambridge, MA) at 1-5 x 105 viable cells/ml, 0.1 ml/well in Eagle's minimum essential medium (EMEM) (BioWhittaker, Walkerville, MD). Cells were incubated for 24 hrs at 37 ~ C in 5% C02 to obtain monolayers, the MuIFN ~/~ (4 x 10 6 international units fU/mg protein) (Lee Biomolecular, San Diego, CA), concentration range 8 x 10-8 x 10 4 units/ml (U/ml) was added and cell cultures incubated for the next 24-72 hrs. The morphology of the treated cells and controls were observed throughout the incubation time using a light microscope. At the end of incubation time (24 or 72 hrs) the proliferation and viability of cells were evaluated using 3H-thymidine incorporation and MTT assay, respectively. The 3H-thymidine (specific activity 6.7 Ci/mmol, New England Nuclear NEN, Wilmington, DE) was added to each well at 5 gCi/ml for 4 hrs following IFN treatment of cells. The incorporation of 3Hthymidine into cells DNA was measured as counts per minute (CPM) using Beckman scintillation counter as previously described [17]. Data were obtained from 6 replicate cultures in three experiments. In parallel cultures cell viability was evaluated using the MTT assay based on reduction of MTT [(4,5-dimethylthiazol-2yl)-2,5-diphenyl-tetrazoliumbromide] (Sigma Chemical Co., St. Louis, MO) by reducing enzymes present only in viable, metabolically active cells according to Mosmann [18] with modifications [19, 20]. The absorbance of formazan was measured at 540 nm using an Emax microplate reader (Molecular Devices, Menlo Park, CA). The amount of formazan formed by viable cells was expressed as optical density of 6 replicate cultures in three experiments. The MTI" assay was previously found [19] to reflect glioma cell viability (formazan level corresponded to the cell number).

G A G isolation To obtain GAG, the G-26 cells were seeded at 2 x

11 106 cells in 75 cm 2polystyrene flasks and cultured in DMEM supplemented with 10% fetal bovine serum, (2 mM) L-glutamine, (50 U/ml) penicillin, and (50 gg/ml) streptomycin at 37 ~ C in the presence of 5 % CO 2. The following day, the cultures were divided into controls (non-treated) and MuIFN ~/~3 treated at 8 x 102 or 8 x 103 U/ml (based on the obtained proliferation/viability dose response curve for MuIFN ~/[3) and incubated for 72 hrs. At the end of incubation, two fractions: the incubation medium and the cells were collected from every tissue culture flask. The incubation medium was centrifuged at 1000 x g for 10 min to remove floating cells and the remaining supernatant fluid was used for G A G extraction. The cell fraction was obtained by adding 5 ml of ice cold 50 mM Tris buffered 0.9% NaC1 (pH 7.5) to every tissue culture flask (previously washed twice with the same buffer) and cells were scraped with a rubber 'policeman' and sonicated for 60 sec. Each fraction, (the medium or the cells) obtained from two 75 cm 2tissue culture flasks were pooled together totaling approximately 4 x 106 cells in 10 ml of Tris buffered 0.9% NaC1 (cell fraction) or 30 ml of medium and then used for GAG isolation. A total of 6 experiments were performed containing 2-4 replicates of samples in both control and IFN treated group. Both the culture medium and glioma cell suspension were digested for 16 hrs at 50 ~ C with protease type XXV (Pronase E) (Sigma Chemical Co) at 15 mg/ml [21]. At the end of digestion both fractions were centrifuged at 10,000 x g for 10 min and the pellet was discarded. The supernatants from both fractions were treated with 0.5 % cetylpyridinium chloride (Sigma Chemical Co.) and left for 16-24 hrs at 26 ~ C to precipitate the GAG [22]. Next, the precipitate from each fraction was centrifuged at 30,000 x g for 10 min, resuspended in one volume of ice cold 2 M NaC1 and four volumes of ethanol (total of 35 ml) and then centrifuged at 10,000 x g for 10 min. The precipitated GAG was then washed once in 10 ml of 100% alcohol (absolute alcohol) pelleted at 10,000 • g for 10 rain and finally dissolved in i ml of distilled water. Next, the preparation of G A G was concentrated by oven evaporation at 60 ~ C for 24-48 hrs. Finally, the dried GAG samples were resuspended in 200 gl of distilled water.

Determination of GAG levels The uronic acid levels as GAG equivalents were measured in fractions of culture medium and G-26 cells using glucuronolactone as the standard according to the method of Bitter and Muir [23]. The presence of uronic acid in G A G preparations was measured spectrophotometrically at 530 nm as previously described [22] and its levels were used to express the amount of GAG produced by glioma cells in vitro [24]. The evaluation of uronic acid levels was based on uronic acid reaction with carbazole using the experimental samples or various concentrations ( 4 4 0 gg/ml) of standard glucuronolactone (Sigma Chemical Co.). The levels of uronic acid in the GAG samples from both fractions (the cells and supernatant), were expressed in gg/ml based upon a standard curve. In each experiment, all samples were adjusted to contain the same level of uronic acid (0.2-1 gg depending on the GAG yield) and used for further evaluation using cellulose acetate electrophoresis.

Enzymatic digestion of GAGs To identify purified GAGs, the experimental samples and standards were enzymatically digested with the specific degrading enzymes and evaluated using cellulose acetate electrophoresis as described previously [25]. The aliquots (10-50 gl) of isolated GAG fractions (0.2-0.7 gg GAG) or standards (0.8-1 gg GAG) were mixed with equal volumes of enzyme solution and digested at 37 ~ C. The enzyme solutions were prepared as follows: Leech hyaluronidase (825 U/ml) in 0.1 M sodium acetate/0.1 M NaC1 pH 5.4 used at 10,000 U/ml (1 hr digestion); Streptomyces hyalurolytic hyaluronidase (10,000 U/ml) in 0.1 M sodium acetate/0.1 M NaC1 pH 5.4 (4 hrs digestion); heparinase 2 mg/ml in 0.1 M sodium acetate buffer (1 hr digestion); keratinase 0.5 U/ml in 0.1 M Tris-acetate pH 7.3 (3 hrs digestion); chondroitinase ABC (0.55 U/rag), used as 5 U/ml in 0.1 M Tris-HC1 pH 8.0 (1 hr digestion); chondro-4-sulfatase CSA (30 U/mg protein) and/or chondro-6-sulfatase CSC (5.6 U/rag protein) both in 50 mM Tris-buffer pH 8.0 used at 0.5 units (3 hrs

12

Fig. 1. Top Panel: The celluloseacetate electrophoresisof GAG isolated from a) G-26 cell culture medium b) G-26 cells. Two bands, HA and CS, are visiblein the GAG preparation obtained from cultured medium. Bottom Panel: The cellulose acetate electrophoresisof GAG sample (S) and standard chondroitin sulfates: CSA (A), CSB (B), CSC (C). Sample and standards were either non-treated or treated with chondroitinaseABC (E). The chondroitinaseABC treatment of standards CSA + E = (AE), CSB + E = (BE), CSC + E = (CE) and the sample S + E = (SE) resulted in their digestion.The treatment of sample (S) with streptomyces hyalurolytic hyaluronidaseS + H = (SH) resulted in the digestion of the top HA band.

digestion). All enzymes and standards were purchased from Sigma Chemical Co.

from 6 experiments (total of 13-14 samples) (experiments using MuIFN c~/[3treatment). Statistical analysis by Student's t-test was used to compare control (n = 13) and IFN treated (n = 14) groups.

Cellulose acetate electrophoresis The individual G A G s were identified using cellulose acetate electrophoresis according to Schmidley and Blue [22]. The isolated, concentrated experimental and standard G A G samples (enzymatically digested or not) were normalized to contain equivalent amounts of G A G (0.2-1 gg) and 25 gl samples were applied using a Gelman applicator onto Gelman-Super Sepraphore membranes. Electrophoresis was performed on Gelman Semi-micro electrophoresis system with 0.2 M calcium acetate buffer at p H 7.0, and separation of G A G s was achieved in 20 min at 10 mA/membrane constant current. Glycosaminoglycan bands were visualized by staining the membrane with 0.1% Alcian blue/acetic acid for 5 min and destained with multiple washes in 10% acetic acid. The quantitative analysis of G A G bands was performed densitometrically. The visage program of the Bio-Image video scanning analysis system (Millipore Corp., Bedford, MA) was used for densitometric analysis and the data were expressed as integrated optical density (IOD). The evaluated samples were obtained from either 3 experiments (total of 6-9 samples) (experiments identifying the type of CS by digestion with CSA or CSC enzymes) or

Results

The analysis by cellulose acetate electrophoresis of G A G s isolated from G-26 cells and incubation medium showed that the majority of G A G s were secreted extracellularly into the cell culture medium (Fig. i a, b top panel). Intracellular G A G s consisted of a small fraction of the total isolated G A G . The electrophoretic pattern of G A G s isolated from the culture medium and the cells demonstrated two bands: the top broad band and the lower narrow band (Fig. 1A). To identify G A G species produced by G-26 glioma cells the enzymatic digestion of the G A G s from isolated culture medium and cells was performed. However, data presented are based on the G A G derived from supernatant since levels of cellular G A G were not adequate. It was found that neither digestion with heparinase nor keratinase affected the G A G bands. However, following treatment with Leech hyaluronidase (which is known to degrade H A and CS) both bands were digested. To further identify G A G species, the preparations were treated with highly purified Streptomyces hyalurolytic hyaluronidase (specifically degrades HA), resulting in the disappearance of the top band, con-

13 A 0 cJ

20

. o~

I-

-Jr" 9- .

m I=

9

0

B

2

U --

3. o

1

0

80

800

O000

SO000

MulFN-alpha/beta(U/ml) Fig. 3. The effect of 72 brs incubation with MuIFN ~/13 on G-26 (A) proliferation and (B) viability. Proliferation was evaluated by 3H-thymidine incorporation measured as counts per minute (CPM). Viability was evaluated by MTI" assay measured as optical density at 540 nm. Presented are means and standard deviations obtained from 12 replicates in two experiments.

Fig. 2. Top Panel: Cellulose acetate electrophoresis of standard chondroitin sulfate A (A) and G A G sample (S). The treatment of the standard (A) and the sample (S) with the specific chondroitinase A (a) resulted in digestion of the standard (Aa) but not sample (Sa). Bottom Panel: Cellulose acetate electrophoresis of standard chondroitin sulfate C (C) and G A G sample (S). The treatment of the standard (C) and the sample (S) with the specific chondroitinase (c) resulted in digestion of the standard (Cc) but not sample (Sc).

firming that it was H A (Fig. 1, Bottom Panel). To identify the bottom G A G band, a series of digestions with chondroitinase ABC (degrades both chontroitin sulfates ABC as well as HA) (Fig. 1, Bottom Panel) and chondro-4 sulfatase (digests CSA) and chondro-6-sulfatase (digests CSC) were performed (Fig. 2). It was found that chondroitinase ABC digested both the top band (already identified as HA) and the bottom band, (Fig. 1, Bottom Panel) confirming that the bottom band was CS. To identify the type of CS produced by the G-26 cells, the specific degrading enzymes of CSA or

CSC were used. However, this treatment did not affect the lower G A G band (Fig. 2) (also confirmed by quantitative densitometric evaluation of CS (data not shown)). It was found that 72 hrs incubation of G-26 cells with various concentrations (8 x 10-8 x 104 U/ml) of MuIFN cz/[3resulted in a dose dependent inhibition of G-26 cell proliferation and viability (Fig. 3A, B). Following 24 hrs incubation, a moderate inhibitory response was observed which was similar for both proliferation (measured by 3H-thymidine incorporation) and viability (measured by MTT assay), e.g. 10-20% inhibition at 8 x 102 U/ml and 30% at 8 x 103 U/ml of MuIFN c~/13(data not shown). A higher inhibitory effect on proliferation/viability was observed after 72 hrs incubation with MuIFN c~/138 x 102 U/ml. For example, incubation with MuIFN cx/[3 8 x 102 U/ml resulted in approximately 30% inhibition of proliferation and 22 % inhibition of viability. Following 72 hrs incubation with MuIFN cz/~3at 8 x 103 U/ml approximately 75 % inhibition of cell proliferation, and 45 % reduction of viability was observed (Fig. 3A, B). There were no visible changes

14

Fig. 4. Cellulose acetate electrophoresis of G A G samples from control G-26 cells n = 3 (lane 1, 2, 3) and MuIFN ot/~ 72 hrs treated G-26 cells n = 3 (lane 4, 5, 6). The top band of G A G is H A and the bottom band is CS. The G A G sample from IFN treated cells shows decreased levels of CS.

in cell morphology when observed microscopically following incubation with MulFN 0U[~at both concentration, 8 x 102 and 8 x 103 U/ml. Based on the above observations, incubation of glioma cells for 72 hrs with MuIFN cU[~at 8 x 102 and 8 x 10~U/ml was used to evaluate GAG production by these cells. Treatment with the lower IFN concentration had no significant effect on GAG levels. However, the higher concentration resulted in a significantly decreased secretion of CS when evaluated by cellulose acetate electrophoresis (Fig. 4). The quantitative densitometric evaluation of the GAG bands separated by cellulose acetate electrophoresis from non-treated and IFN treated cells (13-14 samples from six experiments), showed no effect on 0.10 m co r

I m

o_ 0

0.05

e r

0.00

N

Control

-T~n=14N MulFN

Fig. 5. Densitometric evaluation of CS separated using cellulose acetate electrophoresis from non-treated and 72 hrs MuIFN cU[3 treated G-26 cells. Presented are means and standard deviations of integrated optical density from 13-14 samples. The decreased level of CS in IFN treated compared to control cells is statistically significant, p < 0.008.

HA levels. However, a significant (p < 0.008) inhibition of CS production was observed in MuIFN o~/~ treated cells (IFN treated cells: 0.03 + 0.007 IOD; control cells: 0.07 + 0.01 IOD) (Fig. 5).

Discussion

The effect of MulFN cU[~on synthesis of GAG by mouse glioma cells is evaluated in this study. Two types of GAGs, HA and CS, are found to be secreted extracellularly by glioma G-26. The data in Fig. 2 led to the conclusion that the bottom band was neither CSA nor CSC. The lower band may be a different type of CS such as CSB (dermatan sulfate) or one of the CS isomers such as D, E, E G and H for which currently there are no commercially available specific digesting enzymes. However, all types of CS and its isomers can be digested by chondroitinase A B e [9]. As previously shown, the elevated levels of proteoglycans/glycosaminoglycans (PGs/GAGs) are produced by glioma in vitro [8, 24]. In vivo GAG contents of glial tumors also differ both qualitatively and quantitatively from the GAG levels found in normal white matter [7]. For example, low grade astrocytomas show a three fold higher concentration of total GAGs (due to elevated HA, CS and HS) than normal white matter. The high levels of HA and CS are found in the cortex infiltrated by astrocytoma, implicating GAG's role in tumor invasiveness [11]. Also the presence of HA, CS and HS in the blood vessels of glioma is correlated with vessels'

15 hyalinization, thickening and deformation, possibly affecting BBB permeability within the glioma. The elevated level of CS in brain vessels also is correlated with grade of glioma [8]. It was shown previously that tumor formation depends on the biosynthesis of PGs/GAGs [6]. Under normal conditions, the sulfate groups of polysaccharide chains of G A G allow for PGs interaction with soluble ligands and matrix proteins involved in cell adhesion. However, neoplastic transformation of cells results in alteration of PGs synthesis and especially of their polysaccharide chains both in the tumor and surrounding tissues. These changes often result in decreased sulfation of some of the GAGs. The tumor tissue of a number of malignancies is known to contain significantly elevated levels of CS when compared to corresponding normal tissue [6, 8, 27]. These alterations in synthesis of PGs/ GAGs are believed to stimulate tumorigenic growth by decreasing the adhesion of transformed cells to the extracellular matrix [6]. The pattern of PGs/GAG changes varies among tumors of different histological types [6]. This also includes various tumors of the CNS in which qualitative differences in G A G content between astrocytomas, glioblastomas and oligodendrogliomas are observed [7, 11]. For example, in gliomas it is suggested that increased histological grade is associated with either an increase [8] or decrease [7] in CS levels. IFN also is reported to inhibit proliferation of various tumor cells in vitro [28] including glioma cells [29, 30]. In general, IFN sensitivity of malignant cells, e.g., glioma, may vary among cells of numerous established cell lines [30]. In this study it was observed that 72 hrs incubation of G-26 cells with various concentrations of MuIFN ~/13 affected their proliferation/viability in a dose dependent manner as previously observed by others using human IFN a and ~ in human glioma studies [30]. This laboratory observed earlier that incubation of G-26 cells for 24 hrs with MulFN c~/[3affected their proliferation/viability and the ICs0 value was approximately 3.6 x 10 4 U/ml [26]. It was shown by Numa et al. [30], that human IFNs (a and 13) at I x 104 U/ml reduced the number of human glioma cells by 50%. Also in this study, 8 x 103 U/ml showed approximately 50 % decreased viability of G-26 glioma. The

higher MuIFN o~/~concentration, e.g., 8 x 10 4 U/ml which were below ICs0 values led to reduced proliferation of G-26 in vitro and cell death as shown in Fig. 3A, B. The MTT assay used for evaluation of cell viability was found previously to directly correlate to cell number in glioma culture [19]. It previously has been suggested that the effect of IFN on cell proliferation may relate to its modulatory activity on the cell's size, shape, cytoskeleton and fibronectin distribution, as well as cell membrane organization, rigidity, motility and function [12-16]. There is only a limited number of reports evaluating IFN (mainly IFN-y) effects on production of G A G by fibroblasts, synovial or endothelial ceils in models associated with tissue damage and inflammation, e.g., scleroderma [13], sepsis [14], and rheumatoid arthritis [15]. However, IFN ~ and 13effects on G A G production in vitro are seldom evaluated. The only data available comparing IFN O~A,C~Dand ~3 to IFN 7 in human lung fibroblast cells [10] show that these ct and [3IFNs in contrast to IFN 7 did not stimulate GAG. While inhibition of glioma cell proliferation by IFN is reported [29, 30], the effect of IFN on G A G levels in glioma is not. The only report of IFN effects on glioma cell cytoskeleton relates to an in vitro model using rat tumor cells. Incubation of rat glioma cells with rat IFN (type not specified) resulted in morphological changes of cells and a decrease in fibronectin produced by the glioma cells [16]. The authors conclude that IFN effects on the glioma's cell cytoskeleton may be responsible for the observed antiproliferative effect and could lead to redifferentiation of cells towards a non-transformed phenotype. In this study we showed that prolonged treatment (72 hrs) of mouse G-26 cells with MuIFN c~/~3 resulted in decreased cell proliferation/viability which correlated with significantly decreased extracellular CS levels. This is consistent with a previous report that shows that CS, which is elevated in colon carcinoma, is degraded by enzymes which also act as in vitro growth inhibitors of these cells [27, 31]. Therefore, the observed in vitro effect of MuIFN ct/[3 on CS levels in glioma may underlie IFN induced inhibition of glioma proliferation. The inhibitory effect of MuIFN c~/13on CS induction may be explained by recent findings that IFN y

16 binds to the G A G part of basement membrane PG [32]. The sulfated groups of G A G were identified as essential for this recognition and unsulfated polysaccharides such as H A did not have the capacity to bind IFN 5,. It was suggested that IFN y in vivo may be binding to basement membrane which would then store IFN ,/ around the cells, protecting it from proteolytic degradation and loss of activity [32]. Our data show that only CS, but not HA, levels were significantly affected by MuIFN (z/J3. Therefore, it is possible that MuIFN o~/[~binds to CS, reducing its production. Furthermore, CS inhibition by IFN may result in slowing down glioma invasiveness in vivo. It has been suggested that the in vivo inhibition of GAGs synthesis, e.g., H A and CS may correlate with decreased invasiveness of the tumor through lowering direct and indirect interactions with host cells [33]. Recently, some of the GAGs in glioma, such as hyaluronate (HA binding protein) or heparin sulfate, were identified as cell-adhesion molecules along with the highly glycosylated CD-44 molecules differentially expressed on CNS tumors [34-36]. The CD-44 adhesion molecule (identical to lymphocytehoming receptor) was recently identified as a receptor for H A and CS [35, 36]. It also has been shown that these adhesion molecules can mediate cell attachment to extracellular matrix components [34]. Their differential expression may play a role in tumor-cell migration and local invasiveness in glioma. The modification of these adhesion molecules, e.g., through IFN treatment, may inhibit glioma progression. The observed effect of MuIFN a/[3 on G A G secretion by glioma in vitro may be the underlying event in inhibition of glioma progression in vivo. It was observed in vivo in this laboratory that MuIFN o~/[3delivered, following osmotic BBB alteration at the early stage of G-26 progression (3 days post intracerebral implant) significantly extended animal survival time [17]. Therefore, further evaluation of MulFN a/J3 effect on G-26 glioma's G A G will be studied in the in vivo G-26 intracerebral glioma model to define whether the in vivo inhibition of glioma growth by IFN could be related to its effect on GAG.

Acknowledgements This project was supported by grant R01 NS28989 awarded by the National Institute of Neurological Disorders and Stroke. The authors wish to thank Mildred Acevedo-Duncan, Ph.D. for performing cellulose acetate electrophoresis; Kathy Cho Roetzheim and Ken Olejar for their excellent technical support; Cheri Chance and Pat Urban for their secretarial assistance and Steve Specter, Ph.D. for critical review of this manuscript.

References 1. Groothuis DR, Vick NA: Brain tumors and the blood-brain barrier. Trends Neurosci 5: 232-235, 1982 2. Neuwelt EA, Barnett PA: Blood-brain barrier disruption in the treatment of brain tumors: Animal studies. In: Implications of the blood-brain barrier and its manipulation. (ed) E.A. Neuwelt. Plenum Publ Corp, New York and London, 1989, Volume 2, pp 107-182 3. McBride WH, Bard JBL: Hyaluronidase-sensitive halos around adherent cells. J Exp Med 149: 507-515, 1979 4. Gately CL, Messler Muul L, Greenwood MA, Papazoglou S, Dick SJ, Kornblith PL, Smith BH, Gately MK: In vitro studies on the cell-mediated immune response to human brain tumors. II. Leukocyte-induced coats of glycosaminoglycan increase the resistance of glioma cells to cellular immune attack. J Immuno1133: 3387-3395, 1984 5. Bourin M-C, Lindahl U: Glycosaminoglycans and the regulation of blood coagulation. Biochem J 289: 313-330,1993 6. Esko JD, Rostand KS, Weinke JL: Tumor formation dependent on proteoglycan biosynthesis. Science 241: 10921096, 1988 7. Bertolotto A, Magrassi ML, Orsi L, Sitia C, Schiller D: Glycosaminoglycan changes in human gliomas. A biochemical study. J Neuro-Oncol 4: 43-48, 1986 8. Schrappe M, Klier FG, Spiro RC, Waltz TA, Reisfeld RA, Gladson CL: Correlation of chondroitin sulfate proteoglycan expression on proliferating brain capillary endothelial cells with malignant phenotype of astroglial cells. Cancer Res 51: 4896-4993, 1991 9. Murata K, Yokoyama Y: Enzymatic analysis with chondrosulfatases of constituent disaccharides of sulfated chondroitin sulfate and dermatan sulfate isomers by high-performance liquid chromatography. Anal Biochem 149: 261-268, 1985 10. Elias JA, Krol RC, Freundlich B, Sampson PM: Regulation of human lung fibroblast glycosaminoglycan production by recombinant interferons, tumor necrosis factor, and lymphotoxin. Clin Invest 81: 325-333, 1988 11. Giordana MT, Bertolotto A, Mauro A, Migheli A, Pezzotta

17

12.

13.

14.

15.

16. 17.

18.

19.

20.

21.

22.

23. 24.

S, Racagni G, Schiffer D: Glycosaminoglycans in human cerebral tumors. Part II. Histochemical Findings and Correlations. Acta Neuropatho157: 299-305, 1982 Pfeffer LM, Wang E, Tamm I: Interferon effects on microfilament organization, cellular fibronectin distribution, and cell motility in human fibroblasts. J Cell Biol 85: 9-17, 1980 Gillery R Serpier H, Polette M, Bellon G, Clavel C, Wegrowski Y, Birembaut R Kalis B, Cariou R, Maquart FX: Gamma-interferon inhibits extracellular matrix synthesis and remodeling in collagen lattice cultures of normal and scleroderma skin fibroblasts. Eur J Cell Biol 57: 244-253, 1992 Klein NJ, Shennan GI, Heyderman RS, Levin M: Alteration in glycosaminoglycan metabolism and surface charge on human umbilical vein endothelial cells isolated by cytokines, endotoxin and neutrophils. J Cell Sci 102: 821-832, 1992 Daireaux M, Redini F, Loyau G, Pujol J-P: Effects of associated cytokines (IL-1, TNF-c~, IFN-7and TGF-[3) on collagen and glycosaminoglycan production by cultured human synovial cells. Int J Tiss Reac 12: 21-31, 1990 Stratton MR, Martin JM, Lantos PL: The effects of interferon on glial cells. Br J Exp Path 68: 643-653, 1987 Wiranowska M, Gonzalvo AA, Saporta S, Gonzalez OR, Prockop LD: Evaluation of blood-brain barrier permeability and the effect of interferon in mouse glioma model. J Neuro-Onco114: 225-236, 1992 Mosmann T: Rapid colorimetric assay for cellular growth and survival: Application to proliferation and cytotoxicity assays. J Immunol Meth 65: 55-63, 1983 Nikkhah G, Tonn JC, Hoffman O, Kraemer HR Darling JL, Schonmayr R, Schachenmayr W: The MTT assay for chemosensitivity testing of human tumors of the central nervous system Part 1: Evaluation of test-specific variables. J NeuroOnco113: 1-11,1992 Sobottka SB, Berger MR: Assessment of antineoplastic agents by MTT assay: Partial underestimation of antiproliferative properties. Cancer Chemother Pharmacol 30: 385393, 1992 Kennedy A, Frank RN, Mancini MA: In vitro production of glycosaminoglycans by retinal microvessel cells and lens epithelium. Invest Ophthalmol Vis Sci 27: 746-754, 1986 Schmidley JW, Blue P: Isolation of glycosaminoglycans from basement membranes of brain microvessels. Brain Res Bull 20: 27-31, 1988 Bitter T, Muir HM: A modified uronic acid carbazole reaction. Anal Biochem 4: 330-334, 1962 Steek PA, Moser RR Bruner JM, Liang L, Freidman AN,

25.

26.

27.

28.

29. 30.

31. 32.

33.

34.

35.

36.

Hwang T, Alfred Yung WK: Altered expression and distribution of heparan sulfate proteoglycans in human gliomas. Cancer Res 49: 2096-2103, 1989 Kanwar YS, Farquhar MG: Isolation of glycosaminoglycans (heparan sulfate) from glomerular basement membranes. Proc Natl Acad Sci USA 76: 4493-4497, 1979 Naidu AK, Wiranowska M, Kori SH, Roetzheim KC, Kulkarni AP: Inhibition of cell proliferation and glutathion Stransferase by ascorbyl esters and interferon in mouse glioma. J Neuro-Onco116: 1-10, 1993 Gowda DC, Bhavanandan VP, Davidson EA: Isolation and characterization of proteoglycans secreated by normal and malignant human mammary epithelial cells. J Biol Chem 11: 49264934, 1986 Baron S, Tyring SK, Fteishmann R, Coppenhaver DH, Niesel DW, Klimpel GR, Stanton J, Hughes TK: The interferons: Mechanisms of action and clinical application. J Amer Med Assoc 266: 1375-1383,1991 Lundblad D, Lungren F: Block of glioma cell line in S by interferon. Int J Cancer 27: 74%754, 1981 Numa Y, Kawamoto K, Sakai N, Matsumura H: Flow cytometric analysis of antineoplastic effects of interferon-~,[3 and labelled with flourescein isothiocyanate on cultured brain tumors. J Neuro-Oncol t1: 225-234, 1991 Takeuchi J: Effect of chondroitinases on the growth of solid Ehrlich ascities tumor. Br J Cancer 26: 115-119, 1972 Lortat-Jacob H, Grimaud JA: Binding of interferon-gamma to heparan sulfate is restricted to the heparin-like domains and involves carboxylic - but not N-sulfated - groups. Biochim Biophys Acta 1117: 126-130, 1992 Toole BR Biswas C, Gross J: Hyaluronate and invasiveness of the rabbit V2 carcinoma. Proc Natl Acad Sci USA 76: 6299-6303, 1979 Kuppner MC, Van Meir E, Gauthier Th, Hamou MF, de Tribolet N: Differential expression of the CD44 molecule in human brain tumors. Int J Cancer 50: 572-577, 1992 Aruffo A, Stamenkovic I, Melnick M, Underhill CB, Seed B: CD44 is the principal cell surface receptor for hyaluronate. Cell 61: 1303-1313,1990 C~ WT' de Trib~ N' Antel JR Gauthier T' Kuppner MC: Adhesion molecules and malignant gliomas: implications for tumorigenesis. J Neurosurg 76: 782-791, 1992

Address for offprints: Marzenna Wiranowska, University of South Florida, College of Medicine, Department of Neurology, MDC Box 55, 12901 Bruce B. Downs Blvd. Tampa, FL 33612, USA