Expression and Significance of UrokinaseType Plasminogen Activator Gene in

Human Brain Gliomas

Xiang Zhang, MD, PhD,* Zhou Fei, MD, PhD, Xing-yao Bu, MD, PhD, Hai-ning Zhen, MD,Zhi-wen Zhang, MD, PhD, Jian-wen Gu, MD, AND Yi-jun Chen, MD

Neurosurgical Research Institute, Xi-Jing Hospital, The Fourth Military Medical University,Xian, P.R. China

Background and Objectives: Urokinase type plasminogen activator(uPA) regulates a variety of processes involved in tissue morphogenesis,cell differentiation, migration and invasion. We analyzed the availableinformations to better interpret the pathogenetic relationship between uPAactivity and the malignant biological behavior of human brain gliomas.Methods: We retrospectively studied the presence and distribution of uPAin human brain gliomas by Northern blot hybridization and immunohis-tochemical methods in 43 cases of brain gliomas and 5 cases of normalbrain tissues.Results:All tissues expressed 2.5 kb transcripts of uPA mRNA. The uPAmRNA levels were significantly higher in high-grade gliomas than inlow-grade gliomas and normal brain tissues (P < 0.01). Levels of uPAmRNA expression in tumor tissues with recurrence in 18 months postop-eratively and survival period less than 3 years were significantly higherthan counterparts (P < 0.01). The distribution of uPA protein in the im-munoreactivity was mainly in tumor cells and microvascular endothelialcells of glioblastomas and anaplastic astrocytomas, localizing at cyto-plasms, especially near sites of vascular proliferation and at the leadingedges of tumors.Conclusions:High expression of uPA gene is associated with the malig-nant progression of gliomas and demonstrates a high level of correlationwith the recurrence and invasive behaviors of high grade gliomas.J. Surg. Oncol. 2000;74:90–94. © 2000 Wiley-Liss, Inc.

KEY WORDS: brain glioma; plasminogen activator; gene

INTRODUCTION

The discovery that tumor cells can synthesize uroki-nase type plasminogen activator (uPA) was firstly madeby Astedt et al. [1]. The uPA activity is closely related tothe malignant biological behavior of tumors [2,3]. Recentstudies have shown that human uPA gene is located at thechromosome 10 and comprises 11 exons, and it is a 54kdenzyme shown to participate in the tissue degradationunder certain normal and pathological conditions, includ-ing cancer invasion and metastasis [4]. Increased uPAexpression has been found in cancers of the breast, lung,colon and prostate, as well as correlated with the worseoutcomes in patients with lung or breast cancer. The

uPA-mediated fibrinolytic system plays a pivotal role inthe process of invasion and metastasis in human malig-nant tumors [5].

In the present study we used Northern blot hybridiza-tion and immunohistochemical methods to study the ex-pression of uPA and analyze its clinical significance inhuman brain gliomas.

*Correspondence to: Xiang Zhang, MD, PhD, Prof. & Dir. of Neuro-surgical Research Institute, Xi-Jing Hospital, Fourth Military MedicalUniversity, No.15, West Changle Road, Xian 710032, P.R. China.E-mail: xzhang@fmmu.edu.cnAccepted 27 March 2000

Journal of Surgical Oncology 2000;74:90–94

© 2000 Wiley-Liss, Inc.

MATERIALS AND METHODS

Forty-three human brain glioma and 5 normal braintissue (NB) specimens were collected from patients whowere admitted to the hospital. For Northern blot hybrid-ization, the samples were flash-frozen in liquid nitrogenimmediately after surgical removal. For immunohisto-chemistry of uPA protein and haematoxylin and eosinstaining, the samples were embedded in paraffin and sec-tioned at a thickness of 4mm. According to the histo-logical typing criteria of the central nervous system tu-mors, the samples included 18 low-grade gliomas (LGG,Grade I and II), 25 high-grade gliomas (HGG, Grade IIIand IV), among them including 11 anaplastic astrocyto-mas (AA, Grade III) and 14 glioblastomas multiforme(GBM, Grade IV). There are no oligo cases in the presentstudy. Patients had not been subjected to chemotherapyor radiotherapy preoperatively. They underwent total re-moval or partial removal with curative intent for the firsttime, and followed with postoperative adjuvant chemo-therapy or radiotherapy. The patients included 24 menand 19 women, with ages ranging from 3–72 years (me-dian, 42 years). The follow-up period was longer than 3years, 22 cases died from tumor recurrence within 3 post-operative years and among them, 15 cases recurred in 18postoperative months.

Northern Blot Hybridization

A 1.5-kilobase uPA cDNA probe was used (a gift fromProfessor B.J. Beatrice, University of Texas MD, Ander-son Cancer Center, Houston, TX).b-Actin was used asan internal control. The method was performed as de-scribed previously [6]. Briefly, total RNA from gliomaand normal brain tissues was extracted with the use of astandard guanidinium thiocyanatephenol chloroformtechnique. Twentymg of total RNA was electrophoresedin 1.5% agarose gels and transferred to Nyotran-modifiednylon filters by capillary action. Cloned cDNA probeswere labeled with a random primed labeling kit (Boeh-ringer-Mannheim, Germany) using [32P]dCTP as radio-active label. The filters were then hybridized overnight at42°C with the prehybridization solution containing theradiolabeled and denatured 1.5-kb cDNA specific foruPA mRNA. The filters were then washed in 1× SSC-0.75% sodium dodecyl sulfate at 50°C and dried. X-rayfilms were exposed to the filter for 1–3 days at −70°Cusing intensifying screens. RNA loading equalities werechecked by reprobing the blot with a cDNA correspond-ing to glyceraldehyde phosphate dehydrogenase. Thelevels of uPA mRNA were determined by densitometryand normalized to the control probe.

Microvessel Quantity

The immunohistochemical LSAB method wasadopted. The factor VIII-related antigen (FVIIIRAg)

monoclonal antibody and LSAB kit were all purchasedfrom DAKO Corp. (Carpinteria, CA). The microvesselquantity (MVQ) in tumor tissues was quantified accord-ing to Maeda’s method [7]. Briefly, any endothelial cellor endothelial cell cluster positive for factor VIII-relatedantigen and clearly separate from an adjacent cluster wasconsidered to be a single, countable microvessel (MV), 5high magnification (×200) fields per section were se-lected in areas of the most intense neovascularization,counted, and represented the mean ± SE.

uPA Immunohistochemistry

The routine immunohistochemical ABC method wasused, referring to the instructions of the kit (DAKOCorp.).

Statistical Analysis

Student’s t-test, linear correlation analysis, and theuPA mRNA linear correlation with MVQ variable wereused.

RESULTSNorthern Blot Analysis of uPA

All tissues expressed 2.5 kb transcripts of uPA mRNA.The uPA mRNA was found to localize predominantly tothe leading tumor edge, whereas uPAR mRNA was ex-pressed throughout the tumor and its level was signifi-cantly higher in HGG than they were in LGG and NB (P< 0.01) (Fig.1). Expression of uPA mRNA was barelydetectable in low-grade astrocytomas and NB, and nosignificant difference was observed between them (P >0.05). Levels were significantly higher in AA group andGBM group than they were in LGG group and AA group,respectively (P < 0.01). Levels of uPA mRNA expres-

Fig. 1. Analysis of mRNA expression of urokinase type plasminogenactivator gene in gliomas,b-actin as internal control. Lane 1: normalbrain tissue; lane 2: glioma (Grade III); lane 3: glioma (Grade IV);lanes 4,5: glioma (Grade I); lanes 6,7: glioma (Grade II).

Plasminogen Activator Gene in Gliomas 91

sion in tumor tissues with recurrence in 18 months post-operatively and a survival period less than 3 years wassignificantly higher than counterparts (P < 0.01) (TableI). The uPA mRNA was also expressed in tumor cellsnear necrotic areas. mRNA expression was considerablycorrelated with the MVQ in gliomas (r4 0.56; P <0.01), but had no significant correlation with gender orage of the patients and tumor size.

Immunohistochemical Detection of uPA

The uPA protein expression was consistent with theuPA mRNA expression. The distribution of uPA proteinin the immunoreactivity was mainly in tumor cells andmicrovascular endothelial cells of glioblastomas and AA,localizing at cytoplasms, especially near sites of vascularproliferation and at the leading edges of tumors. It wasalmost undetectable in LGG and NB (Fig. 2).

MVQ of Tumor Tissues

MV showed brown-yellow staining and presented het-erogeneous distribution in different grades of glioma tis-sues. The MVQ in AA and GBM groups were markedlyhigher than in LGG group (P < 0.01 between AA andLGG; and P < 0.002 between GBM and LGG). TheMVQ in cerebral tissues with light, middle and heavydegree peritumoral edema were also significantly morethan in cerebral tissues without peritumoral edema (P <0.01, orP < 0.005). The MVQ in recurrence group (50.4± 23.9) was significantly more than in non-recurrencegroup (29.1 ± 13.4;P < 0.01) (Table II).

DISCUSSION

It has been shown that proteolytic degradation of theextracellular matrix and stromal elements is a crucialevent in the process of cancer invasion and metastasis.uPA is one type of series proteases that is secreted bytumor cells and other cells and occurs as a single-chaininactive proenzyme form (pro-uPA). Binding of pro-uPAto its specific receptor (uPA-R) on the tumor cell surfaceaccelerates the conversion of pro-uPA to two-chain ac-tive enzyme form and consistent with the mRNA detec-tion, uPA-R and uPA proteins were expressed by malig-nant brain tumors [8]. Increased levels of uPA-R havebeen found in primary malignant brain tumor tissues,especially highly malignant glioblastomas, and, to alesser degree, in malignant astrocytomas, suggesting thatthis receptor might be involved in efficient activation ofpro-uPA and confinement of uPA activity on the cellsurface of invading brain tumors [9,10]. The active uPAis capable of converting plasminogen into the active pro-teolytic enzyme plasmin, the latter can lead to extracel-lular matrix and stromal elements degradation and pro-teolysis. By this, uPA participates in and regulates theprocess of tumor cell migration and tumor invasion

TABLE I. The Relationship Between Urokinase TypePlasminogen Activator mRNA Expression and PathologicalGrade and Prognosis in Gliomas

Group NuPA mRNA

(x ± s)

Normal brain (NB) 5 2.54 ± 1.20Low-grade glioma (LGG) 18 2.68 ± 2.14a

Anaplastic astrocytoma (AA) 11 14.83 ± 5.76b

Glioblastoma multiforme (GBM) 14 47.39 ± 13.58c

Recurrence (18 postoperative months)Yes 15 38.59 ± 14.62No 28 7.02 ± 6.67d

Survival period (years)<3 22 35.55 ± 15.21>3 21 4.42 ± 6.67e

aCompared with NB group (P > 0.05).bCompared with LGG group (P < 0.01).cCompared with AA group (P < 0.01).dCompared with group with recurrence (P < 0.01).eCompared with group with survival period less than 3 years (P <0.01).

Fig. 2. Expression of urokinase type plasminogen activator in ana-plastic astrocytoma, the positive products were localized at cytoplasmsof tumor cells (ABC method, ×200).

TABLE II. The Relationship Between Microvessel Quantity(MVQ) and Histological Grade and Recurrence in Gliomas

Group NMVQ(x ± s)

Normal brain (NB) 5 24.3 ± 11.2Low-grade glioma (LGG) 18 26.3 ± 10.8Anaplastic astrocytoma (AA) 11 45.1 ± 17.6Glioblastoma multiforme (GBM) 14 57.8 ± 20.2Recurrence 15 50.4 ± 23.9Non-recurrence 28 29.1 ± 13.4

92 Zhang et al.

[4,5,11]. Recent reports also indicate that uPA gene andits protein product were overexpressed in several humanneoplasms including gastric carcinoma, breast carcinomaand lung carcinoma, and associated with progression,differentiation, invasion, metastasis and prognosis of theabove tumors [2,5,12].

With regard to the relationship between uPA and glio-mas, Moonen et al. [13] have shown that purified humanuPA induced the proliferation and outgrowth of glial fi-brillary acidic protein-positive cells from newborn ratcerebellar microexplants, and concluded that uPA had amitogenesis for astrocytes. In immunofluorescence stud-ies, uPA has been localized to cytoplasms of culturedglioblastoma cells. Human glioma cell lines that pro-duced uPA grew more rapidly than those did not. Inaddition, uPA showed positive immunostaining in reac-tive astrocytes in the brain areas adjacent to or infiltratedby tumor [1,4]. Yamamoto et al. [14] have shown that theuPA secreted by glioma cells provides a crucial cellularmechanism in initiating the process of tumor invasion.This is the principal reason for failure of all local thera-pies in the treatment of malignant gliomas.

The present studies show that all gliomas and NB canexpress 2.5 kb transcripts of uPA mRNA. Low levels ofuPA mRNA were found in all 5 normal brain tissues and18 LGG tissues, but no significant differences exist be-tween the two groups (P > 0.05). The levels of uPAmRNA elevated markedly with the increase of histopath-ological grade and malignant degree of gliomas. The lev-els of uPA mRNA in AA group is significantly higherthan that in LGG group (P < 0.01). The levels of uPAmRNA in the GBM group is significantly higher thanthat in AA group (P < 0.01). The levels of uPA mRNAin glioma tissues with recurrence within 18 postoperativemonths and a survival period of less than 3 years aresignificantly higher than in glioma tissues without recur-rence within 18 postoperative months and with a survivalperiod longer than 3 years (P < 0.01). The uPA proteinexpression was consistent with uPA mRNA expressionmainly distributed in tumor cells and endothelial cells ofGBM and AA tissues, with only small amounts in LGGtissues. Results indicated that glioma cells can synthesizeand secrete uPA, and the uPA involves in and promotesthe progression in the process of initiation and develop-ment of gliomas. uPA gene expression is associated withthe relapse and prognosis of patients with gliomas.

Angiogenesis has certain invasiveness to tissues. Tu-mor cells are able to invade into peripheral tissuesthrough tissue cracks cleaved by the newborn vascula-ture. The massive angiogenesis with the proliferation ofvascular endothelial cells is one of the malignant pheno-types of gliomas. Although malignant glioma cells arecapable of producing uPA, the vascular cells may be themain resources of endogenous uPA in vivo. The endo-thelial cells at the tips of growing capillaries can secrete

degradative enzymes, that help vessels enter the tumorentity gradually and form new vessel nets in tumor tis-sues. Because newly proliferating capillaries have frag-mented basement membranes and are leaky, this facili-tates the escape of tumor cells into neovasculature,surviving the circulation and proliferating in the targettissues, thereby resulting in the invasion and metastasisof tumors [15,16]. In glioma tissues, several tumor an-giogenesis factors have been found, including fibroblastgrowth factor, vascular endothelial growth factor, plate-let-derived growth factor. These cause not only mitogen-esis of cells but they are also potential inducers of uPAsynthesized by endothelial cells [17,18]. Hsu et al. [4]reported that the proliferating vascular endothelium hadimmunoreactivity in uPA strong positive gliomas,thereby speculating that uPA may play a role in tumorangiogenesis. In this study, the MVQ of gliomas wascounted by immunohistochemical staining of the factorVIII-related antigen using its monoclonal antibody. Itwas found that MVQ was significantly correlated withuPA mRNA expression. In addition, the levels of uPAprotein were elevated in proliferating endothelial cells ofhigh grade gliomas. The results indicate that expressionof uPA gene is associated with the malignant progressionof gliomas, that may play an important role in the recur-rence and invasive behaviors of high-grade gliomas.

REFERENCES

1. Astedt B, Holmberg L: Immunological identity of urokinase andovarian carcinoma plasminogen activator released in tissue cul-ture. Nature 1996;261:595–597.

2. Kinder DH, Berger MS, Mueller BA, et al.: Urokinase plasmin-ogen activator is elevated in human astrocytic gliomas relative tonormal adjacent brain. Oncol Res 1993;5:409–414.

3. Landau BJ, Kwaan HC, Verrusio EN, et al.: Elevated levels ofurokinase-type plasminogen activator and plasminogen activatorinhibitor type-1 in malignant human brain tumors. Cancer Res1994;54:1105–1108.

4. Hsu DW, Efird JT, Hedley Whyte ET: Prognostic role of uroki-nase-type plasminogen activator in human gliomas. Am J Pathol1995;147:114–123.

5. Heiss MM, Babic R, Allgayer H, et al.: Tumor-associated prote-olysis and prognosis: new functional risk factors in gastric cancerdefined by the urokinase-type plasminogen activator system. JClin Oncol 1995;13:2084–2093.

6. Xingyao B, Baixiang J, Yingjiang Y, et al.: Correlation betweennm23-H1 mRNA and the clinical behavior of nephroblastoma.Chin J Clin Oncol 1997;24:179–182.

7. Maeda K, Chung YS, Takatsuka S, et al.: Tumor angiogenesis andtumor cell proliferation as prognostic indicators in gastric carci-noma. Brit J Cancer 1995;72:319–323.

8. Gladson CL, Pijuan Thompson V, Olman MA, et al.: Up-regulation of urokinase and urokinase receptor genes in malignantastrocytoma. Am J Pathol 1995;146:1150–1160.

9. Chintala SK, Mohanam S, Go Y, et al.: Altered in vitro spreadingand cytoskeletal organization in human glioma cells by down-regulation of urokinase receptor. Mol Carcinogen 1997;20:355–365.

10. Mohanam S, Sawaya WE, Yamamoto M, et al.: Proteolysis andinvasiveness of brain tumors: role of urokinase-type plasminogenactivator receptor. J Neurooncol 1994;22:153–160.

11. Bindal AK, Hammoud M, Shi WM, et al.: Prognostic significance

Plasminogen Activator Gene in Gliomas 93

of proteolytic enzymes in human brain tumors. J Neurooncol1994;22:101–110.

12. Reith A, Rucklidge GJ: Invasion of brain tissue by primaryglioma: evidence for the involvement of urokinase-type plasmin-ogen activator as an activator of type VIII collagenase. BiochemBiophys Res Commun 1992;186:348–354.

13. Moonen G, Grau-Wagemans MP, Selak I, et al.: Plasminogenactivator is a mitogen for astrocytes in developing cerebellum.Brain Res 1985;352:41–48.

14. Yamamoto M, Sawaya R, Mohanam S, et al.: Expression andlocalization of urokinase-type plasminogen activator in humanastrocytomas in vivo. Cancer Res 1994;54:3656–3661.

15. Bu Xingyao, Zhang Xiang, Chen Yijun, et al.: Research on rela-tionship between the microvessel quantity of astrocytoma and itsbiological behavior. Med J Chin PLA 1999;24:25–27.

16. Takano S, Yoshii Y, Kondo S, et al.: Concentration of vascularendothelial growth factor in the serum and tumor tissue of braintumor patients. Cancer Res 1996;56:2185–2192.

17. Chen ZQ, Fisher RJ, Riggs CW, et al.: Inhibition of vascularendothelial growth factor induced endothelial cell migration byETS1 antisense oligonucleotides. Cancer Res 1997;57:2013–2019.

18. Weidner N: Intratumor microvessel density as a prognostic factorin cancer. Am J Pathol 1995;147:9–17.

94 Zhang et al.