IM

LJIC

R

sdcmetoftwttbogTbfHspbo

m

omstsvm

lm

Biochemical and Biophysical Research Communications 273, 411–416 (2000)

doi:10.1006/bbrc.2000.2952, available online at http://www.idealibrary.com on

nfluence of bcl-2-Related Proteins on Matrixetalloproteinase Expression in a Rat Glioma Cell Line

isa Oliver, Karine Tremblais, Nathalie Guriec, Stephane Martin,* Khaled Meflah,ean Menanteau, and Francoise-Marie ValletteFR 26, U 419 INSERM, 9 quai Moncousu, 44035 Nantes Cedex 01, France; andlinique de Neurochrirugie, CHU, Nantes, France

eceived May 22, 2000

to the intrinsic chemo- and radio-resistance of thesethttduipM(aapatbl

imr(ocew7ipte1cktitl

The expression of bcl-2-related proteins has beenhown to be a key element in tumoral malignancy. Theegradation of the extracellular matrix (ECM) by spe-ialized matrix metalloproteinases (MMPs) is anotherajor step in tumor invasion and metastasis. We have

xamined, in a rat glioma cell line A15A5, the effect ofhe stable transfection of human bcl-2, bax and bcl-xln MMPs expression. Using a zymographic assay, weound that all transfected cell lines expressed a gela-inase activity which is predominantly associatedith MMP-9. In bcl-2 and bcl-xl transfected cells, the

ranscription of MMP-9 was decreased compared tohat of control or bax transfected cells. In addition, inax transfected A15A5, we observed a down regulationf TIMP-1, the inhibitor of MMP-9. These results sug-est that the ratio between MMP-9 and its inhibitorIMP-1 is tightly controlled in cells overexpressingcl-2 related proteins (i.e., high ratio in bax trans-ected A15A5 and low ratio in bcl-2 transfected A15A5).owever, MMPs secreted by bcl-2 transfected cells were

till capable of hydrolyzing FasL present on human lym-hocytes. Our results suggest that the expression ofcl-2 related proteins could participate in the regulationf MMP-9/TIMP-1 in gliomas. © 2000 Academic Press

Key Words: apoptosis; bcl-2 related proteins; gliomas;etalloproteinases.

The major reason for the therapeutic failure in gli-ma treatment is their poor response to adjuvant che-otherapy or radiotherapy. In addition, the rapid

preading and subsequent seeding of glioma cells fromhe original site of the tumor impairs the efficacies ofurgical resection (1). The resistance of gliomas to con-entional therapies is not only due to the difficulty ofost drugs to cross the brain-blood barrier but also due

Abbreviations used: CNS, central nervous system; ECM, extracel-ular matrix; FasL, Fas ligand; LDH, lactate dehydrogenase; MMP,

atrix metalloprotease; TIMP, inhibitor of MMP.

411

umors (2). The expression of antiapoptotic moleculesas been shown to be an important obstacle in conven-ional anticancer treatments such as radio- or chemo-herapies of tumors (3). Apoptosis or programmed celleath is the process used by metazoans to eliminatennecessary or potentially harmful cells. This program

s tightly controlled by a family of proto-oncogenes therototype of which is the product of the BCL-2 gene.embers of this family can be either anti-apoptotic

e.g., bcl-2, bcl-xl) or pro-apoptotic (e.g., bax, bid . . .)nd can form homor- or heterodimers (4). Pro- andntiapoptotic bcl-2 related proteins are often coex-ressed within the same cell and the ratio between pro-nd antiapoptotic proteins is a major determinant ofhe fate of the cell. Dysregulation of apoptotic control iselieved to be a major step in tumor progression, ateast, in the final steps of malignant transformation (5).

Degradation of the ECM by MMPs appears to benvolved in tumor invasion and metastasis. Experi-

ental evidence has directly implicated MMPs in theemodeling of the stromal tissue surrounding tumors6). These proteases have been implicated in a numberf developmental and pathological processes. MMPsan generate active matrix protein fragments, influ-nce the release and the activation of growth factorshich in turn modulate angiogenesis (see for example), tumor cell growth and apoptosis (8). MMPs are alsonvolved in the processing or shedding of cell surfaceroteins such as CD44 or TNF (8). Among the proteinshought to be the target of MMPs is the protein, FasLxpressed at the surface of cytotoxic lymphocytes (9,0). In the latter case, the shedding of Fas L by MMPsould be an important factor in the prevention of theilling of tumoral cells by cytotoxic T cells. In additiono the action of these proteases, their receptors and/ornhibitors can also be directly involved in cell migra-ion, although their precise role is not clearly estab-ished (8).

0006-291X/00 $35.00Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

Invasive glioma cells migrate locally along the cen-tLh(Mt

e(clae2s

M

uubC(f

w(scaflwm

Dvp

epi2aef6rmwbtcAy

AcPpc

were assayed for protein concentration using the Bradford techniquep

pbbeptV(

ctC

pTrp5wGtbfi1cfpau

R

T

tutvneeftmCsecmtnttbwUA

Vol. 273, No. 2, 2000 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

ral nervous system (CNS) white matter fiber tracts.ow grade cerebral gliomas are infiltrative whereasigh grade malignant gliomas are extremely invasive11). Several studies have pointed out an increase in

MPs expression during the malignant transforma-ion of gliomas (see for example 12).

A link between bcl-2 related proteins and MMPsxpression has been established recently by Wick et al.13). These authors showed that endogenous bcl-2ould enhance migration and invasion of human ma-ignant gliomas by altering the expression of MMPsnd their inhibitors. In this study, we have studied thenzymatic activity and RNAs expression of MMP 9 andas well as TIMP-1 in the rat glioma cell line A15A5

tably transfected with bcl-2, bax and bcl-xl.

ATERIALS AND METHODS

Materials and reagents. Unless otherwise specified, all reagentssed in this study were from Sigma (France). Commercial antibodiessed were raised against human bax (Immunotech, France), humancl-2 (Dako, France); anti-human FasL and bcl-xl were from Santaruz (Tebu, France). MMP inhibitor (KB8301) was from Pharmingen

France Biochem, France) and the cocktail of protease inhibitors wasrom Boehringer Ingelheim (France).

Cell culture and transfections. The rat glioma cell line, A15A5,as obtained from the European Collection of Animal Cell Culture

Valbonne, France). The cell line was maintained in RPMI-1640upplemented with 10% heat-inactivated FCS, 100 mg/ml streptomy-in, 100 units/ml penicillin and 2 mM L-glutamine in a 5% CO2

ir-humidified atmosphere at 37°C. Lymphocytes were obtainedrom donors from the Blood Transfusion Center of Nantes and iso-ated on a ficoll density gradient by centrifugation. The cells wereashed twice in RPMI-1640 prior to the addition of conditionededia from transfected A15A5.Plasmids encoding for human bcl-2, bax or bcl-xl were gifts fromr. G. Evan (ICRF, London) and were subcloned into pRcCMV (In-itrogen, USA). The construct baxTMXL used, was constructed asreviously described (14).For transfection experiments, A15A5 cells were transfected with

ither 2 mg control pRcCMV vector, pRcCMV bcl-2, pRcCMVbcl-xl,RcCMV bax, or pRcCMV baxTMxl. Plasmid DNAs were introducednto 106 cells by electroporation (GenePulser, BioRad, France) using00 V/cm and 250 mF. Transfected cells were selected and cloned inmedium containing neomycin (250 mg/ml). Two different clones

ncoding equivalent amounts of either bax, bcl-xl or bcl-2 were usedor these experiments. Stable transfected cells were cultivated to0–80% confluency and washed several times in RPMI in order toemove traces of serum. Cells were then maintained in serum-freeedium for 24 h then aliquots of the Cell Conditioned Media (CCM)ere assayed for MMP assay (Zymograms) or FasL digestion (seeelow). Apoptosis was induced in some experiments by a brief UVreatment as described before; and both control and UV treated wereultured for a further 24 h in the serum free medium conditions.poptosis was assayed by morphological and caspase activity anal-sis as described before (14).

Cell extracts and Western blots. Lymphocytes (7.5 3 105 cells) or15A5 (105 cells) were homogenized vol/vol in RIPA buffer (PBSontaining 1% NP-40, 0.5% Na-deoxycholate, 0.1% SDS, 10 nMMSF, 10 nM aprotinin, 1 nM Na-orthovanadate). After severalassages in a 2 ml glass Dounce homogenizer, the homogenates wereentrifuged at 4°C at 13,000g for 30 min. The resulting supernatants

412

rior to analysis on 15% SDS–PAGE.Western blots were performed as described by Juin et al. (15) using

rimary antibodies anti-bcl-2 (1 mg/ml), anti-bcl-xl (0.25 mg/ml), anti-ax (2 mg/ml), anti-FasL (2 mg/ml) and actin (0.5 mg/ml). The anti-odies bound to Immobilon-P (Millipore, France) were detected bynhanced chemiluminescence (Amersham, France) using a seconderoxydase-labeled antibody). The amount of immunoreactive pro-eins was quantified using IPLab Gel Program (Signal Analytics,ienna, USA) after scanning with an Appligene-Oncor Imager

Strasbourg, France).

Gelatinase A and B/MMP-2 and -9 activities. CCM from differentell cultures were obtained after 24 h culture under serum-free condi-ions as described above and zymograms were performed with 50 mlCM using gelatin as a MMP substrate as previously described (16).

RNA preparation and RT-PCR analysis. Total RNA was pre-ared using a SV Total RNA Isolation System (Promega, France).he cDNA was prepared from 1 mg total RNA using Super Scripteverse transcriptase (GIBCO-BRL, France). The oligonucleotiderimers used for the amplification of rat gelatinase B were sense9-ataacgagttctctggcg-39 and antisense 59-aggaggtcataggtcacg-39hich resulted in an amplified fragment of 231 bp and for ratAPDH were sense 59-catcaccatcttccaggagc and antisense 59-

tctgagtggcagtgatggc which resulted in an amplified fragment of 341p. The primers were synthesized by Genosys (UK). For the ampli-cation: 40 ng cDNA was amplified in a final volume of 30 ml usingU Taq polymerase (GIBCO-BRL, France). 30 amplification cycles

onsisting of 95°C for 30 s; 55°C for 30 s; 72°C for 30 s were per-ormed. The number of cycles were within the linear phase of am-lification of the RNAs analyzed. 10 ml of the amplified products werenalyzed in a 2.5% agarose/TBE gel. The bands were quantifiedsing the IPLab Gel Program.

ESULTS

ransfection of bcl-2-Related Proteinsin A15A5: Sensitivity to Apoptosis

The rat glioma cell line, A15A5, was transfected withhe human bcl-2, bcl-xl, bax or baxTMXL as describednder Materials and Methods. It should be noted thathe endogenous levels of rat bcl-2, bcl-xl and bax wereery low in the control pRcCMV-transfected cells (dataot shown). For each bcl-2 related protein, two clonesxpressing similar amounts of transfected proteins andquivalent doubling times were selected. The trans-ected cells were grown to 60–80% confluency. Half ofhe culture dishes were subjected to a short UV treat-ent (UVB, 1 min) in order to induce apoptosis (14).ells were then incubated for an additional 24 h undererum-free conditions. Sensitivity to apoptosis wasvaluated both by morphological criteria (chromatinondensation, cellular blebbing . . .) and biochemicalodifications such as the induction of caspase activi-

ies or the release of the cytosolic lactate dehydroge-ase (LDH). The phenotypes induced by the transfec-ions were consistent with the nature of the proteinsransfected: antiapoptotic molecules such as bcl-2 andcl-xl delayed the onset of the apoptotic processhereas bax, a proapoptotic protein potentiated theV induced apoptosis. No potentiation was observed in15A5 cells transfected with the construct baxTMxl in

wbcfictacrttfoli

M

u

cells. Cells were cultivated for 24 h in absence of serumii1twtrt(twpfsctgtigteiaTarRctssobb3apb

gMch(

a

Vol. 273, No. 2, 2000 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

hich the C-terminus of bax was substituted by that ofcl-xl. Both apoptotic or control cells were assayed foraspase-3 like activity which is representative of thenal executive phase of apoptosis in UV-treated A15A5ells. As illustrated in Fig. 1, we used the cleavage ofhe fluorogenic peptide ac-DEVD-amc in a whole cellssay as a quantitative index of caspase activity. Wehose to illustrate the caspase activities at 24 h whichepresents the peak of DEVDase activity in bax-ransfected cells (data not shown). As shown in Fig. 1,he level of DEVDase activity was higher in bax trans-ected cells than in pRcCMV-transfected cells. On thether hand, the expression of DEVDase activity wasower in bcl-2, bcl-xl or baxTMxl transfected cells thann pRcCMV-transfected cells.

MP-9/2 Activities in Transfected Cells

Zymograms were performed as described earlier (16)sing control or UV-treated CCMs from transfected

FIG. 1. A caspase-3 like activity was assayed using the fluoro-enic substrate of this caspase (ac-DEVDamc) as described underaterials and Methods. Caspase activity was assayed in transfected

ells (pRcCMV, human bax, the construct baxTMxl, human bcl-2 oruman bcl-xl) 24 h after UV treatment (UV) or in untreated cellscont). This illustration is representative of 3 different experiments.

FIG. 2. Zymograms of CCM derived from A15A5 cells transfecte10% SDS–PAGE as described under Materials and Methods. This

413

n order to avoid inhibition by a2-macroglobulin. Sim-lar volumes (25 ml) were loaded for each condition on0% PAGE impregnated with 0.5% gelatin. As illus-rated in Fig. 2, two types of gelatinase/MMP activitiesere observed in CCMs, one major band corresponding

o MMP-9 (92 kDa) and a minor band at 72 kDa cor-esponding to MMP-2. Transfected cells, regardless ofhe nature of the plasmid, secreted gelatinase activityFig. 2), the MMP-9 activity being always higher thanhat of MMP-2 (Fig. 2). The relative MMP-9 activityas decreased in bcl-2 and bcl-xl transfected cells com-ared to the activity secreted by pRcCMV or bax trans-ected cells (Fig. 2). In apoptotic cells, although the nonpecific release of cytosolic enzymes such as LDH wasonsiderably increased, very little gelatinase/MMP ac-ivity was found in CCM (Fig. 2). Similarly, little or noelatinase/MMP activity was detected in cellular ex-racts from the cells, suggesting that most of the activ-ty was secreted (data not shown). These results sug-est that expression of anti apoptotic protein reducedhe amount of MMP-9 secreted in A15A5 cells. We thusvaluated the level of transcription of MMP-9 and itsnhibitor TIMP-1, by performing semi-quantitativenalyses of mRNA expression by RT-PCR. RT-PCR forIMP-1 and MMP-9 were carried out, using GAPDH asn internal control, after determination of the linearange of amplification of the different mRNAs. TotalNA extracted from bax, bcl-2 and bcl-xl transfectedells were used as well as RNAs from A15A5 cellsransfected with pRcCMV or the chimeric bax con-truct (BaxTMxl). The RT-PCR was performed as de-cribed in Materials and Methods. The RNA expressionf MMP-9 was found to be higher in pRcCMV, bax oraxTMxl transfected cells compared to that in bcl-2 orcl-xl transfected cells. The RT-PCR illustrated in Fig.was representative of three different experiments

nd the ratio between MMP-9 vs GAPDH mRNAs ex-ression was 0.6 6 0.08 for pRcCMV; 0.67 6 0.12 forax; 0.59 6 0.20 for baxTMxl; 0.24 6 0.6 for bcl-xl and

ith human bcl-2, bax, bcl-xl or pRcCMV. 25 ml CCM was loaded onstration is representative of 3 different experiments.

d willu

0etMdpb

ervwceT

3R

Vol. 273, No. 2, 2000 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

.22 6 0.16 for bcl-2. Nonsignificant differences in thexpression of MMP-9 mRNAs were found between cellsransfected with pRcCMV or bax (P 5 0.4478) or baxT-xl (P 5 0.9448) (Fig. 3B). In contrast, significant

ifferences were found between A1A5 transfected withRcCMV and cells expressing bcl-2 (P 5 0.0212) orcl-xl (P 5 0.0032). This result suggested that the

FIG. 3. RT-PCR analysis of GAPDH, MMP-9 and TIMP-1 RNA de, baxTMxl; 4, bcl-2; 5, bcl-xl. This illustration is representative ofT-PCR.

414

xpression of mRNAs coding for MMP-9 was down-egulated in bcl-2 and bcl-xl transfected cells. Con-ersely, we found that TIMP-1, the inhibitor of MMP-9as down-regulated in bax and baxTMxl transfected

ells whereas bcl-2, bcl-xl or pRcCMV transfected cellsxhibited similar levels of TIMP-1 mRNAs (Fig. 3).hese results suggest thus that the expression of

ed from A15A5 cells transfected. (A) Agarose gel: 1, pRcCMV; 2, bax;ifferent experiments. (B) Quantification of fragments amplified in

riv4 d

Mtb

M

syHMa(ieBMt1Clsipiasc

D

ttrttba

sion of apoptotic molecules such as bax and bcl-2 ap-pltpMefetlpaataagwebudtlnepsolottaiOmrptMewstc(TlwawaIli

pie

Vol. 273, No. 2, 2000 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

MP-9 and TIMP-1 is adversely regulated in cellsransfected with pro- or antiapoptotic members of thecl-2 family.

MP-Mediated Digestion of FasL

We have studied the in situ activity of the MMPsecreted by transfected cells by analyzing the proteol-sis of a MMP-9 substrate: fas ligand (FasL) (9, 10).uman lymphocytes were obtained as described underaterials and Methods. Cell extracts were obtained

fter homogenization in RIPA buffer and homogenates25 mg) were incubated with 25 ml CCMs (100 mg) andncubated at 37°C for 3 h. Aliquots were treated withither the general antiprotease inhibitors cocktail fromoehringer Mannheim (without EDTA) or with theMP inhibitor BK8301. At the end of the incubation,

he samples were heat-inactivated and analyzed in a5% SDS–PAGE followed by immunoblotting for FasL.CMs from all transfected cells were capable of hydro-

yzing FasL from its membrane bound form (p40) to itsecreted form (p27) as previously shown (9, 10). Asllustrated in Fig. 4 for pRcCMV and bcl-2 CCMs, thisroteolysis was inhibited by BK8301 but not by generalnhibitor cocktail. The latter result suggests that FasLt the surface of lymphocytes can be proteolysed to aimilar extent in CCMs of control and bcl-2 transfectedells.

ISCUSSION

Gliomas and especially glioblastoma multiforme (as-rocytoma grade IV, according to the WHO classifica-ion) are usually radio- and chemoresistant and diffuseapidly within brain tissue rendering antitumoralreatments and surgical resection often quite inopera-ive (1). The resistance to anticancer treatments coulde due, at least partially, to the dysfunctioning of thepoptotic program in these tumors. Indeed, the expres-

FIG. 4. FasL digestion by CCM of A15A5 cells transfected withRcCMV or bcl-2 in absence (2) or in presence (1) of 100 mM MMPsnhibitor KB8301. This illustration is representative of 3 differentxperiments.

415

ears to be highly regulated in the passage betweenow to high grade in these tumors (17, 18). The infil-rative rather than invasive properties of gliomas ap-ears to be linked to the expression of MMPs (8). TheseMPs are two of the proteases most frequently over-

xpressed in cancer and angiogenesis and are essentialor cell migration or tumor invasiveness (20). Lampertt al. (12) have shown that the expression of MMP andheir inhibitors TIMP was closely related during ma-ignant progression in gliomas. Thus the ratio betweenoison/counter poison (i.e., MMP-9 and TIMP-1 or bcl-2nd bax) appears to be important for the progression ofstrocytomas. Only recently a relationship betweenhese two processes has been established when Wick etl. (13) proposed that bcl-2, in addition to resistance topoptosis, enhanced the migration/invasive capacity oflioma cells by increasing their levels of MMPs. In thisork, we have tried to establish a link between thexpression of pro- and antiapoptotic proteins of thecl-2 family and the expression of MMPs in glioma. Wesed a rat glioblastoma model, the cell line A15A5,erived from N-ethyl-N-nitrosourea-induced glioma inhe rat BD IX. This tumor has been shown to expressarge amounts of plasminogen activator compared toormal glia (19). We analyzed in this cell line theffects of stable transfections with either anti-apoptoticroteins such as bcl-2 or bcl-xl or proapoptotic proteinuch as bax (Fig. 1) on the expression and the secretionf MMP-9 and MMP-2 (Fig. 2). We did not find a corre-ation between the expression of bax and the secretionf active MMP-9/-2 (Fig. 2). However, bcl-2 and bcl-xlransfected cell lines exhibited less gelatinase activityhan the control pRcCMV-transfected cells. RT-PCRnalyses revealed a down-regulation of the mRNA cod-ng for MMP-9 in the bcl-2 and bcl-xl transfected cells.n the contrary, bax transfected cells expressed lessRNAs coding for TIMP-1. Interestingly, this down-

egulation is not associated with bax proapoptoticroperties as it was also observed with a “loss of func-ion” mutant of bax (baxTMxl). The down regulation ofMP-9/2 in bcl-2 and bcl-xl transfected cells is appar-

ntly paradoxical as a decrease in MMP-9 activityould diminish the aggressivity of the tumors. Indeed,

everal groups have reported an increase of TIMPs inumors not only in gliomas (12) but also in colorectalancers (21), gastric carcinomas (22) and breast cancer23). Recently, it has been shown that TIMP-1 but notIMP-2 inhibits the induction of apoptosis in Burkitt’s

ymphoma cell lines and that its expression is coupledith an increase of the expression of bcl-xl (24). Theseuthors found that TIMP-1 suppression of apoptosisas not due to MMP inhibition, as the reduction andlkylation of the TIMP-1 did not abolish this activity.n breast cancer cell lines, Li et al. (25) found a similarink between the expression bcl-2 and that of TIMP-1,ndependently of cell anchorage. The significance of the

down-regulation of MMP-9 in bcl-2 transfected cellsabrcaa(gtefainttlba

osm

A

C

R

Inoue, S., Yoshino, K., Okumura, K., and Yagita, H. (1995)

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

Vol. 273, No. 2, 2000 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

nd the down regulation of its inhibitor TIMP-1 inax transfected cells remains to be defined. However,esistance to apoptosis due to MMPs could also beorrelated with the capacity of these cells to elicit thenti-tumoral immune response by inhibiting the pro-poptotic signal carried by cytotoxic lymphocytes FasL10). Indeed, we have recently observed that therowth of bcl-2 and bax transfected A15A5 was effec-ively under the control of the immune system (Martint al., manuscript in preparation). One could postulaterom our results that TIMP-1 could provide protectiongainst secreted MMP-9 originating from the tumorsn bcl-2 overexpressing cells. Hydrolysis of FasL fromatural killer lymphocytes would diminish the induc-ion of apoptosis through the Fas system while theumor would still be able to induce apoptosis in theymphocytes. Thus a link between the expression ofcl-2 and a high TIMP-1/MMP9 ratio would provide andditional selective advantage for gliomas cells.In conclusion, our study supports a new adverse role

f bcl-2-related proteins on the control of the expres-ion of MMP-9/TIMP-1 which could influence the tu-oral progression of gliomas.

CKNOWLEDGMENTS

This study was supported by grants from the Ligue Contre leancer, ARC, and INSERM.

EFERENCES

1. Forsyth, P., and Cairncross, J. (1995) Treatment of malignantglioma in adults. Curr. Opin. Neurol. 8, 414–418.

2. Collins, V. (1998) Gliomas. Cancer Survey 32, 37–51.3. Rudin, C. M., and Thompson, C. B. (1997) Apoptosis and disease:

Regulation and clinical relevance of programmed cell death.Annu. Rev. Med. 48, 267–281.

4. Gross, A. M., McDonnell, J. M., and Korsmeyer, S. J. (1999) Bcl-2family members and the mitochondria in apoptosis. Genes Dev.13, 1899–1911.

5. Lowe, S. W., and Lin, A. W. (2000) Apoptosis in cancer. Carci-nogenesis 21, 485–495.

6. Himmelstein, B. P., Canete, S. R., Bernhard, E. J., Dilks, D. W.,and Weiss, S. J. (1994) Metalloprotinases in tumor progression:The contribution of MMP-9. Inv. Metast. 14, 246–258.

7. Fujimora, M. G. Y., Morita-Fujimora, Y., Massengale, J., Ka-wase, M., and Chan, P. H. (1999) Early appearance of activatedmatrix metalloproteinase-9 and blood-brain barrier disruption inmice after focal cerebral ischemia and reperfusion. Brain Res.842, 92–100.

8. Yong, V. W., Krekoski, C. A., Forsyth, P. A., Bell, R., and Ed-wards, D. R. (1998) Matrix metalloproteinases and diseases ofthe CNS. Trends Neurosci. 21, 75–80.

9. Kayagaki, N., Kawasaki, A., Ebata, T., Ohmoto, H., Ikeda, S.,

416

Metalloproteinase-mediated release of human Fas ligand. J.Exp. Med. 182, 1777–1782.

0. Tanaka, M., Itai, T., Adachi, M., and Nagata, S. (1998) Down-regulation of Fas ligand by shedding. Nature Med. 4, 31–36.

1. Kleishue, P., Burger, P. C., and Scheithauer, B. W. (1993) Thenew WHO classification of brain tumors. Brain Pathol. 3, 255–268.

2. Lampert, K., Machein, U., Regina-Machein, M., Conca, W., Pe-ter, H. H., and Volk, B. (1998) Expression of matrix metallopro-teinase and their tissue inhibitors in human brain tumors.Am. J. Pathol. 153, 429–437.

3. Wick, W., Wagner, S., Kerkau, S., Dichgans, J., Tonn, J., andWeller, M. (1998) BCL-2 promotes migration and invasiveness ofhuman glioma cells. FEBS Lett. 440, 419–424.

4. Tremblais, K., Oliver, L., Juin, P., Le Cabellec, M. T., Meflah, K.,and Vallette, F. M. (1999) The C-terminus of bax is not a mem-brane addressing/anchoring signal. Biochem. Biophys. Res. Com-mun. 260, 582–591.

5. Juin, P., Pelletier, M., Oliver, L., Tremblais, K., Gregoire, M.,Meflah, K., and Vallette, F. M. (1998) Induction of caspase-3activity by calcium in normal cytosolic extracts triggers nuclearapoptosis in a cell free system. J. Biol. Chem. 273, 17559–17564.

6. Segain, J.-P., Harb, J., Gregoire, M., Meflah, K., and Menanteau,J. (1996) Induction of fibroblast gelatinase B expression by directcontact with cell lines derived from primary tumor but not frommetastases. Cancer Res. 56, 5506–5512.

7. Krajewski, S., Krajewska, M., Ehrman, J., Sikorska, M., Lach,B., Chatten, J., and Reed, J. C. (1997) Immunohistochemicalanalysis of Bcl-2, Bcl-Xl, Mcl-1 and Bax in tumors of central andperipheral nervous system origin. Am. J. Pathol. 150, 805–814.

8. Newcomb, E. A., Bhalla, S. K., Parrish, C. L., Hayes, R. L.,Cohen, H., and Miller, D. C. (1997) Bcl-2 protein expression inastrocytomas in relation to patient survival and p53 gene status.Acta Neuropathol. 94, 369–375.

9. Hince, T. A., and Roscoe, J. P. (1980) Differences in pattern andlevel of plasminogen activator production between a cloned cellline from an ethylnitrosourea-induced glioma and one from nor-mal adult rat brain. J. Cell Physiol. 104, 199–207.

0. Murphy, G., and Gavrilovic, J. (1999) Proteolysis and cell migra-tion: Creating a path? Curr. Opin. Cell Biol. 11, 614–621.

1. Stetler-Stevenson, W. G., Brown, P. D., Onisto, M., Levy, A. T.,and Liotta, L. A. (1990) Tissue inhibitor of metalloproteinase-2(TIMP-2) mRNA expression in tumor cell lines and in humantumor tissues. J. Biol. Chem. 265, 13933–13938.

2. Mimori, K., Mori, M., Shiraishi, T., Fujie, T., Baba, K., Haragu-chi, M., Abe, R., Ueo, H., and Akiyoshi, T. (1997) Clinical signif-icance of tissue inhibitor expression in gastric carcinoma. Br. J.Cancer 76, 531–536.

3. Fong, K. M., Kida, Y., Zimmerman, P. V., and Smith, P. J. (1996)TIMP-1 and adverse prognosis in non-small lung cancer. Clin.Cancer Res. 2, 1369–1372.

4. Guedez, L., Courtmanch, K., and Stetler-Stevenson, M. (1998)Tissue inhibitor of metalloproteinase (TIMP)-1 induces differen-tiation and an antiapoptotic phenotype in germinal center Bcells. Blood 92, 1342–1349.

5. Li, G., Fridman, R., and Kim, H. R. C. (1999) Tissue inhibitor ofmetalloproteinase-1 inhibitor inhibits apoptosis in human epi-thelial cells. Cancer Res. 59, 6267–6275.