Human Brain Tumors and Plasminogen Activator Inhibitor Type-1 in Malignant Elevated Levels of Urokinase-Type Plasminogen Activator

The plasminogen-plasmin system has been found to modulate neoplastic spread and angiogenesis in tumors outside the central nervous system (CNS), but there have been no quantitative studies on the invasive and vascular tumors of the CNS. Quantitative zymography and enzyme-linked immunosorbent assay were used to determine the amounts of urokinasetype plasminogen activator (u-PA), tissue-type plasminogen activator, and plasminogen activator inhibitors type 1 and type 2 (PAI-I and PAI-2) in benign and malignant primary brain tumors (n 28) as well as nonneoplastic brain (n = 5). u-PA and PAI-I antigen were undetectable in normal brain but significantly elevated in glioblastoma multiforme (u-PA, 2.86 +_ 3.01 ng/mg; PAI-1, 8.19 __5.57 ng/mg; P < 0.001). There was no difference, however, in tissue-type plasminogen activator antigen levels among control, benign, or malignant tissues except for a 4to 7-fold increase in acoustic neuroma. PAI-2 was detected at low levels in 2 of the 33 specimens. These findings indicate that malignancy in primary CNS neoplasms is associated with elevated levels of u-PA and PAI-I, supporting the role of the plasminogen-plasmin system in the pathogenesis of CNS malignancy and as a potential biomarker and therapeutic target. I N T R O D U C T I O N The malignant phenotype of CNS 3 neoplasms is characterized by local invasion (1, 2) and angiogenesis (3, 4); however, the mechanisms regulating these processes in brain tumors is poorly understood. Proteolytic degradation of the extracellular matrix is an essential first step in the invasive processes of neovascularization and tumor cell invasion (5, 6). The plasminogen activators, t-PA and u-PA, catalyze the conversion of plasminogen to plasmin, a serine protease with broad specificity that degrades noncollagenous components of the extracellular matrix and converts the zymogen procollagenase to active collagenase (6). Elevated levels of u-PA and PAI-1 in tumor tissue are powerful and independent predictors of poor prognosis in breast cancer (7-10) and have been correlated with histological degree of malignancy in breast, colon, prostate, skin, and other tumors outside of the CNS (11). Nonspecific PA activity in brain tumor extracts has been correlated with the amount of peritumoral brain edema (12); however, the role of u-PA in human brain tumors has been investigated only in vitro (13, 14). In the present study we have examined the amount and activity of u-PA, t-PA, and PA inhibitors by zymography and ELISA in extracts of nonneoplastic human brain and in brain tumor tissues, and now report that u-PA and PAI-1 are selectively increased in glioblastoma. Received 6/22/93; accepted 12/16/93. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. This work was supported, in part, by grants from Northwestern Memorial Foundation, Illinois Neurofibromatosis Inc., National Institutes of Health (CA57781, CA606177) (to S. B.) and the Veterans Administration Central Office (to H. K.). 2 To whom requests for reprints should be addressed, at Director of Neurosurgical Oncology, Division of Neurological Surgery, Northwestern Memorial Hospital, 233 East Erie St., Suite 500, Chicago, IL 60611-2906. 3 The abbreviations used are: CNS, central nervous system; ELISA, enzyme linked immunosorbent assay; u-PA, urokinase-type plasminogen activator; t-PA, tissue-type plasminogen activator; PAl, plasminogen activator inhibitor. Tissue Preparation. Brain tumor samples were collected at the time of surgery, embedded in O.C.T. Compound (Miles Inc. Elkhart, IN), snap frozen in isopentane and stored at -70~ Frozen sections of each sample were cut, stained with hematoxylin and eosin, and then reviewed so that only well preserved, nonnecrotic, representative tissue was included in the study. Samples were then thawed; rinsed in cold phosphate-buffered saline, pH 7.4; and weighed. Tissue fragments were homogenized in an ice bath with a motor-driven Teflon pestle for 3 min in 1 ml of extraction buffer/50 mg of tissue wet weight (0.075 M potassium acetate, 0.3 M NaC1, 0.1 M L-arginine, 10 m i EDTA, 0.25% Triton X-100, pH 4.2) (15). The homogenates were centrifuged at 12,000 • g for 10 min at 4~ and the supernatants were aliquotted and stored at -70~ Protein concentrations were determined by the method of Bradford (16) using an IgG protein standard (Bio-Rad, Richmond, CA). Plasminogen Activator Determinations. Quantification of antigen was performed using commercially available ELISA kits (t-PA, Diagnostica Stago, Asnieres-Sur-Seine, France; u-PA, PAl-l, PAI-2, Biopool, Umea, Sweden). All determinations were made in duplicate on separate ELISA plates. The u-PA ELISA detects u-PA, pro-u-PA, and u-PA-PAI complexes. Zymography was carried out as a modification of the method of DePetro et aL (17). Equivalent protein amounts of tissue extract were electrophoresed under nonreducing conditions on a 10% sodium-dodecyl-sulfate polyacrylamide gel. Each gel included lanes loaded with two-chain t-PA (Mr 68,000; American Diagnostica, Greenwich, CT) and varying amounts of human high-molecular weight u-PA (Mr 54,000; American Diagnostica) to quantify u-PA activity in the tissue extracts. The gels were washed twice for 30 min in 0.1 M Tris:2.5% Triton X-100, pH 8.1 and overlaid on a 10 • 10 cm casein-agarose gel made with 10 ml of Bio-Rad protease gel substrate tablets dissolved in distilled water supplemented with 10 mg/ml of nonfat dry milk and 20 txg/ml of affinity-purified human plasminogen (18). The zymograms were incubated overnight at 4~ then at 37~ and photographed hourly as the lytic bands developed. Quantification of zymographic activity was done by computer-assisted image analysis on an imaging densitometer (Bio-Rad Model GS-670) (19). A numerical value, representing both size and density, was calculated as the sum of the gray scale units per square millimeter of each lytic band, adjusted by subtraction of background. A standard curve for determination of u-PA activity in the tissues extracts was generated from the four u-PA activity standards included on each gel. As a control, casein-agarose underlay gels were prepared without plasminogen to visualize any intrinsic caseinolytic activity in the tissue extracts. In addition, polyclonal anti-t-PA or anti-u-PA antibodies (American Diagnostica) were incorporated into underlay gels to verify the identity of the lytic bands. Calculations and Statistics. Antigen concentrations of u-PA, t-PA, and PAI-1 were expressed as nanograms of antigen per milligram of total protein. Enzyme activities were expressed as milli-international units (mlU) of u-PA per milligram of total protein. Comparisons between tumor types were made by two-tailed analysis of variance, except in the case of normal brain, which had undetectable u-PA and PAI-1 (and therefore no intragroup variance). Differences were considered significant at P values < 0.05. To determine significance of difference from normal brain, levels of u-PA and PAI-1 in each tumor type were compared to zero by Student's t test. Correlations between u-PA, t-PA, PAI-1, and u-PA activity were determined by a Pearson correlation matrix with confidence levels determined by Bonferroni probabilities. The nonparametric Spearman rank test produced similar correlations with equivalent levels of significance.