Ó Springer-Verlag 2000
J Cancer Res Clin Oncol (2000) 126:145±152
ORIGINAL PAPER
S. Takano á K. Tsuboi á A. Matsumura á Y. Tomono Y. Mitsui á T. Nose
Expression of the angiogenic factor thymidine phosphorylase in human astrocytic tumors
Received: 4 August 1999 / Accepted: 20 October 1999
Abstract Thymidine phosphorylase (TP) has been implicated as a potent angiogenic factor and a prognostic factor in various human solid tumors. We investigated the expression of TP in a series of human astrocytic tumors using immunohistochemistry, enzyme-linked immunosorbent assay, and reverse transcriptase/polymerase chain reaction (RT-PCR) analysis. A total of 63 astrocytic tumors [27 glioblastomas (GBM), 19 anaplastic astrocytomas (AA), 17 low-grade astrocytomas (LGA)] and 5 normal brain tissues were immunohistochemically stained with antibodies to TP, vascular endothelial growth factor (VEGF), p53, MIB-1, and factor-VIII-related antigen. They were also evaluated for the degree of apoptosis by a ApopTag kit. Ten tumors (5 GBM, 2 AA, 3 LGA) and 3 normal brain tissues were evaluated for their expression of VEGF and TP by RTPCR analysis. TP was constantly localized in the cytoplasm of astrocytic tumor cells, less intensely in the cytoplasm of vascular endothelial cells, but not in the normal brain. Some of the TP-positive cells were of macrophage origin, but most positive cells were the tumor cells themselves. Vascular density, MIB-1 positivity, p53 positivity, VEGF expression, and the apoptotic index were signi®cantly higher in the TP-positive tumors than in TP-negative tumors. There was a signi®cant correlation between TP and VEGF mRNA expression. In a limited number of glioblastoma cases, the apoptotic index was signi®cantly higher in TP-positive glioblasS. Takano (&) á K. Tsuboi á A. Matsumura Y. Tomono á T. Nose Department of Neurosurgery, Institute of Clinical Medicine, University of Tsukuba, 1-1-1 Tennoudai, Tsukuba, Ibaraki 305-0006, Japan e-mail:
[email protected] Tel.: +11-81-298-53-3167 Fax: +11-81-298-53-3214 Y. Mitsui National Institute of Bioscience and Human Technology, Agency of Industrial Science and Technology, Tsukuba, Ibaraki, Japan
tomas than in TP-negative glioblastomas. In human astrocytic tumors, TP was expressed in the tumor, macrophage, and endothelial cells. TP was a potent angiogenic factor closely associated with cell proliferation and tumor apoptosis. Key words Angiogenesis á Apoptosis á Glioma á Thymidine phosphorylase á Vascular endothelial growth factor Abbreviations TP thymidine phosphorylase á GBM glioblastoma á AA anaplastic astrocytoma á LGA lowgrade astrocytoma á VEGF vascular endothelial growth factor á RT-PCR reverse transcriptase/polymerase chain reaction
Introduction The growth of human astrocytic tumors is dependent on angiogenesis. Various angiogenic factors have been demonstrated in human astrocytic tumors [e.g. vascular endothelial growth factor (VEGF), basic ®broblast growth factor (bFGF), interleukin-8 (IL-8)] (Takano et al. 1996). Recently thymidine phosphorylase (TP) has been demonstrated to be a potent angiogenic factor and a prognostic factor in various human solid tumors including breast, bladder, colon, gastric, kidney, and lung non-small-cell cancer (Giatromanolaki et al. 1997; Imazono et al. 1997; Koukourakis et al. 1997; Kubota et al. 1997; O'Brien et al. 1996; Relf et al. 1997; Takebayashi et al. 1996; Tanigawa et al. 1996; Toi et al. 1995). In the case of brain tumors, only one report has demonstrated TP localization in the macrophages of glioblastoma tissues (Nakayama et al. 1994). In this study, the expression of TP in a series of human astrocytic tumors was measured by immunohistochemistry, enzyme-linked immunosorbent assay (ELISA), and reverse transcriptase/polymerase chain reaction (RTPCR) analysis.
146
Materials and methods Tissue preparation A group of 63 patients with astrocytic tumors (glioblastoma 27, anaplastic astrocytoma 19, low-grade astrocytoma 17) were included in the current study. Brain tumor tissues and normal brain tissues (2 cases of tumor-free brain and 3 of brain remote from the tumor) were obtained during routine surgical procedures performed for diagnostic or therapeutic purposes. A part of the tissues was immediately ®xed in 10% phosphate-buered formalin for 48 h, paran-embedded, and used for routine pathological diagnosis and immunohistochemistry. Other parts of the tissues were immediately frozen in liquid nitrogen and stored at )70°C. Informed consent was obtained from all subjects involved in the current study. Antibodies and immunohistochemistry The Dako LSAB kit for mouse and rabbit primary antibody (Dako, Glostrup, Denmark) was used (Takano et al. 1996). Tissue sections were deparanized and incubated with 10% normal goat serum in phosphate-buered saline (PBS) for 20 min. The sections were then incubated with a monoclonal anti-(thymidine phosphorylase) antibody, 654-1 (Nishida et al. 1996; provided by Nippon Roche Research Center), at a dilution of 1/100 (10 lg/ml IgG), a polyclonal anti-VEGF antibody, A-20 (Santa Cruz Biotech. Inc., Calif.) at a dilution of 1/100 (1 lg/ml IgG), and a monoclonal anti(human von-Willebrand factor) antibody (Dako, Glostrup, Denmark) at a dilution of 1/50 (254 lg/ml) in PBS for 60 min at room temperature. In order to stain macrophages in the tissues, the sections were incubated with monoclonal anti-(human macrophage), KP1 and HAM56 antibodies (Dako), at a dilution of 1/100 (3.5 and 1.5 lg/ml IgG respectively). Chromatographically puri®ed mouse IgG and rabbit IgG (Dako) at the same concentration were used as negative controls. The sections were incubated with biotinconjugated goat anti-(mouse Ig) or anti-(rabbit Ig) for 10 min, followed by washing in PBS for 10 min. They were then incubated with peroxidase-conjugated streptavidin solution for 5 min, followed by washing in PBS for 5 min. After staining with freshly prepared aminoethylcarbazole solution for 10 min, followed by washing for 5 min in tap water, the sections were counterstained with hematoxylin and mounted with aqueous mounting medium. The intracellular VEGF and TP immunostaining was assessed separately for tumor and endothelial cells, using a semiquantitative scale: negative (0±20% of cells stained), weak (20%±50% of cells stained), strong (more than 50% of cells stained). The strong staining was evaluated as positive (Giatromanolaki et al. 1997). In a further experiment, glioma tissues were stained with a monoclonal p53 antibody, DO7 (Dako), at a dilution of 1/500 (0.8 lg/ml IgG), using an Envision kit (Dako), and with a monoclonal MIB-1 antibody (Immunotech) using an LSAB kit. The sections were developed with diaminobenzidine and counterstained with hematoxylin. Nuclei positive for p53 and MIB-1 were determined by counting at least 1000 tumor cells (Takano et al. 1997). Histochemical detection of apoptotic cells and determination of apoptotic index Apoptotic cells were revealed by the ApopTag in situ detection kit (Oncor, Gaithersburg, Md.). The staining procedures were modi®ed according to the manufacturer's instructions. Brie¯y, after deparanization and rehydration, the tissues were digested with proteinase K (20 lg/ml in PBS; Wako, Osaka, Japan) for 20 min at room temperature and washed. Slides were then put into 3% H2O2 for 5 min and washed with PBS. Ten minutes after the addition of the equilibration buer, terminal deoxynucleotidyl transferase enzyme was pipetted onto the sections, which were then incubated at 37°C for 1 h. The reaction was stopped by putting sections in stop/ wash buer. After washing, anti-digoxigenin peroxidase was added
to the slides, which were then washed, stained with diaminobenzidine (Dako) substrate, and counterstained with hematoxylin. A specimen known to be positive for apoptotic cells was used as a positive control for subsequent staining. Substitution of terminal deoxynucleotidyl transferase with distilled water was used as a negative control. The apoptotic index was expressed as the ratio of positively staining tumor cells to all tumor cells, given as a percentage in each case. At least ®ve representative areas without necrosis in a section were selected by light microscopy using 40- to 200-fold magni®cation. A minimum of 3000 cells were counted under a 400-fold magni®cation. Positively staining tumor cells with the morphological characteristics of apoptosis were identi®ed by standard criteria, including chromatin condensation, nucleolar disintegration, and formation of crescentic caps of condensed chromatin at the nuclear periphery (Ansari et al. 1993). Tumor vascular density Vascular density was scored using the vasoproliferative component of the microscopic angiogenesis grading system (MAGS) that has been used to quantify angiogenesis in a variety of tumors (Takano et al. 1996). The number of vessels in a 200´ ®eld (1.0 mm2) was measured in microvessel ``hot spots'' (i.e., microscopic areas containing the densest collections of microvessels, as initially identi®ed under low-power magni®cation) under an Olympus microscope, AHBT3 (Olympus, Tokyo, Japan) on tissue sections stained for von Willebrand factor. Vascular density was de®ned by averaging the number of vessels in the three most vascularised areas. RNA isolation and reverse transcription polymerase chain reaction (RT-PCR) Total RNA was extracted from 13 frozen tissues (5 glioblastomas, 2 anaplastic astrocytomas, 3 low-grade astrocytomas, 3 normal brains) using as RNeasy mini-kit (Qiagen GmbH, Germany) according to the supplier's recommended procedure. We performed RT-PCR with the GeneAmp RNA PCR kit (Perkin-Elmer Cetus, Norwalk, Conn.). Brie¯y, 1 lg total RNA was reverse-transcribed by murine leukemia virus reverse transcriptase in the presence of random hexamer, followed by the indicated cycles of the PCR reaction (95°C for 1 min, 55°C for 1 min, and 72°C for 1 min) in the presence of 2 lM TP-speci®c primers (35 cycles), VEGF-speci®c primers (35 cycles), or the b-actin-speci®c primers (20 cycles) as a control. The TP primers were as follows (Usuki et al. 1994): the reverse primer (5¢-AGGTCACCTGTGATGAGTGG-3¢), complementary to positions 1164±1135, and the forward primer (5¢AGGCAGAGGTCACAATGAGG-3¢) corresponding to positions 793±822 (Usuki et al. 1994). The VEGF primers (Weindel et al. 1992) included the reverse primer (5¢-CCTGGTGAGAGATCTGGTTC-3¢) spanning bases 861±842 and the forward primer (5¢TCGGGCCTCCGAAACCATGA-3¢) spanning bases 141±160. The b-actin primers (Ng et al. 1985) included the reverse primer (5¢-GGAGTTGAAGGTAGTTTCGTG-3¢) spanning bases 2429± 2409 and the forward primer (5¢-CGGGAAATCGTGCGTGACAT-3¢) spanning bases 2107±2126. The predicted sizes of the ampli®ed TP and b-actin DNA products were 372 bp and 323 bp respectively. The VEGF primers were chosen because they ampli®ed exons 3±8 and allowed the dierent VEGF splicing variants to be distinguished. PCR products of 530 bp and 662 bp corresponded with VEGF121 VEGF165 respectively. The quanti®cation of these RT-PCR product levels was performed on a Macintosh computer using the public-domain NIH Image program (developed at the U.S. National Institute of Health). Thymidine phosphorylase ELISA Thirteen frozen tissues (6 glioblastomas, 2 anaplastic astrocytomas, 3 low-grade astrocytomas, 2 normal brains) were homogenized as previously described (Nishida et al. 1996). Brie¯y, each sample of tumor and normal brain tissue was homogenized in a tenfold
147 volume of 10 mM TRIS/HCl buer (pH 7.4) containing 15 mM NaCl, 1.5 mM MgCl2, and 50 lM potassium phosphate, and then centrifuged at 10 000g for 15 min. The supernatant was stored at )80°C until used. The protein concentration of the supernatant extracted from the tissues was determined by using a DC protein assay kit (Bio Rad Laboratories, Hercules, Calif.). Levels of TP in these samples were measured by an ELISA method as described previously (Nishida et al. 1996). The levels of TP were expressed as U/mg protein, where 1 U is equivalent to the amount of TP that generates 1 lg 5-¯uorouracil from 5¢-deoxy-5-¯uorouridine in 1 h. Statistical analyses Vascular density, MIB-1 positivity, p53 positivity, apoptosis index and the densitometric value of VEGF, TP, and b-actin were expressed as means SD. Statistically signi®cant dierences between tumor types were determined by a one-way analysis of variance and the Tukey test. Correlation between VEGF and TP expression was determined by a Pearson correlation matrix with the con®dence level determined by Bonferroni probabilities. All P values were two-sided; values were considered statistically signi®cant for P < 0.05.
Results Immunolocalization and concentration of TP and VEGF in glioma tissues TP antigen was detected consistently in both the tumor cells and the endothelial cells lining tumor-associated vessels by immunohistochemical staining of formalin®xed sections of astrocytic tumors. TP antigen was detected in the tumor cells of 17 out of 27 glioblastomas, 11 of 19 anaplastic astrocytomas, and 2 of 17 low-grade astrocytomas. TP antigen was detected in the tumorassociated vessels of 8 out of 27 glioblastomas, 4 of 19 anaplastic astrocytomas, and 1 of 17 low grade astrocytomas (Fig. 1A, B; Table 1). TP antigen was not detected in the normal brains. The TP concentration, measured by ELISA, was signi®cantly higher in glioblastomas and anaplastic astrocytomas (18.4 8.2 U/mg protein and 19.4 15.6 U/mg protein respectively) than in low-grade astrocytomas and normal brains (1.7 0.2 U/mg protein and 2.0 0.6 U/mg protein respectively) (P < 0.05; Fig. 2). To determine the cellular source of TP, TP-positive serial glioblastoma sections were stained with anti-macrophage antibodies, KP1 and HAM56. Some TP-positive cells were KP1and HAM56-positive (Fig. 1C, D). There were several distinct dierences in the tumor variables between TP-positive and TP-negative tumors (Table 1). TP-positive tumors have high MIB-1 positivities (P < 0.05), high vascular densities (P < 0.01), high p53 positivities (P < 0.05), and high apoptotic indices (P < 0.05) compared to TP-negative tumors. VEGF antigen was also detected consistently in both the tumor cells and the endothelial cells lining tumor-associated vessels, as described previously (Takano et al. 1996). The expression of TP and VEGF in tumor cells was matched in 45 of 63 cases (+/+25, +/)5, )/+13, )/)20, v2-test, P = 0.0007). These results demonstrate
the strong linkage between TP and VEGF expression as a marker of the angiogenic phenotype of astrocytic tumors. All astrocytic tumors contain various grades of malignancy. Even in their most malignant form, glioblastomas, TP immunopositivities were various. There was a distinct dierence in apoptotic index between TP-positive and -negative glioblastomas (Table 1, Fig. 1E±H). TP-positive glioblastomas revealed signi®cantly higher apoptotic indices than TP-negative glioblastomas (2.05 0.80, 0.98 0.22 respectively, P < 0.05). However, other tumor variables, including MIB-1, vascular density, p53 positivities, and VEGF expression were did not dier between the two subgroups. TP and VEGF and mRNA expression in glioma tissues RT-PCR was performed to determine whether TP and VEGF would predict the degree of malignancy and angiogenesis in glioma and tissues. The results presented in Fig. 3 show that both TP and VEGF mRNA levels revealed the degree of malignancy of glioma tissues. When standardized by the amount of b-actin in the lane, the relative amounts of TP were 0.82 0.14 in glioblastomas (GBM), 0.68 0.01 in anaplastic astrocytomas (AA), 0.11 0.02 in low-grade astrocytomas (LGA), and 0.03 0.06 in normal brain, and the relative amounts of VEGF165 were 0.81 0.46 in GBM, 0.41 0.23 in AA, 0.17 0.01 in LGA, and 0.06 0.09 in normal brain. There was a signi®cant correlation between the TP mRNA and VEGF mRNA levels in glioma tissues (Fig. 4, r = 0.75, P < 0.01).
Discussion Thymidine phosphorylase as an angiogenic factor in astrocytic tumors TP is a protein with a wide range of activities, including stimulation of DNA synthesis (Sotos et al. 1994) and angiogenesis (Moghaddam and Bicknell 1992; Haraguchi et al. 1994). Higher levels of TP are observed in a variety of human tumors than are found in the adjacent normal tissues, including the stomach, colon and ovary (Takebayashi et al. 1996) and are associated with the tumor characteristics of invasion and malignant potential in bladder cancer (Kubota et al. 1997). In astrocytic tumors, TP expression paralleled the histological malignancy associated with proliferation potentials measured by MIB-1 positivities. The cellular source of TP was demonstrated prominently in tumor cells and macrophages, and occasionally in endothelial cells. Although TP activities in glioblastomas and anaplastic astrocytomas were signi®cantly high, these activities were lower than in other cancers including invasive
148
149 b Fig. 1A±H Immunohistochemistry of human astrocytic tumors. A, C, E, G Glioblastoma stained with thymidine phosphorylase (TP) antibody. B Low-grade astrocytoma stained with TP antibody. D Serial section of glioblastoma in C stained with anti-macrophage antibody, KP1. F Serial section of glioblastoma in E stained with an ApopTag kit. H Serial section of glioblastoma in G stained with an ApopTag kit. In glioblastomas, note the intense immunoreaction for TP both in tumor cells and endothelial cells (A). In low-grade astrocytomas, note no immunoreaction with TP (B). Some of the TP-positive cells (C) are macrophages (D). TP-positive glioblastoma (E) reveals a high apoptotic index (F), while TP-negative glioblastoma (G) reveals a low apoptotic index (H). Original magni®cation ´200
bladder cancer (Kubota et al. 1997) and colorectal, breast, and gastric cancers (Nishida et al. 1996). Interestingly there were signi®cant dierences in other tumor variables between TP-positive and -negative tumors. TPpositive tumors revealed high vascularity, high VEGF expression, high p53 positivities, and high apoptotic indices compared to TP-negative tumors. Direct evidence of a role for TP in angiogenesis comes from the following experiments: (1) transfection of the TP gene into transformed ®broblasts in nude mice results in increased tumor vascularization (Ishikawa et al. 1989), (2) TP stimulates the migration of cultured endothelial cells (Haraguchi et al. 1994) and (3) TP enhances angiogenesis in model systems in vivo (Moghaddam et al. 1995; Ishikawa et al. 1989). In vivo, positive expression of TP was associated with high vascular density in non-small-cell lung cancer (Koukourakis et al. 1997), breast cancer (Toi et al. 1995), renal cell carcinoma (Imazono et al. 1997), gastric carcinoma (Tanigawa et al. 1996), and colorectal carcinomas (Takebayashi et al. 1996). In our study, expression of TP was signi®cantly correlated with the increment of microvascular density and VEGF expression, which is a Table 1 Summary of the results of the immunohistochemical study. TP thymidine phosphorylase. The apoptotic index is the ratio of positively staining tumor cells to all tumor cells, vascular Tumor
MIB-1 (%)
Vascular density
Fig. 2 Thymidine phosphorylase concentration measured by enzymelinked immunosorbent assay in human astrocytic tumors and normal brains. The concentration was signi®cantly higher in glioblastomas (GBM) and anaplastic astrocytomas (AA) than in low-grade astrocytomas (LGA) and normal brains (P < 0.05)
Fig. 3 Reverse trasncriptase/polymerase chain reaction (RT-PCR) analysis for thymidine phosphorylase (TP), vascular epithelial growth factor (VEGF ) and b-actin in human astrocytic tumors and normal brains. M molecular marker, GBM glioblastoma, AA anaplastic astrocytoma, LGA low-grade astrocytoma
density is the average number of vessels in the three most vascularised areas (1.0 mm2) p53 (%)
VEGF-positive (%)
Apoptotic index
Low-grade astrocytoma TP-positive (n = 2) TP-negative (n = 15) Total (n = 17)
2.1 1.6 2.2 1.5 2.2 1.5
21.0 11.3 13.9 9.5 14.8 9.6
3.3 4.6 7.0 12.9 6.5 12.2
50 26.70 29.40
0.95 0.07 0.84 0.69 0.85 0.63
Anaplastic astrocytoma TP-positive (n = 11) TP-negative (n = 8) Total (n = 19)
11.1 6.8 10.2 8.3 10.7 7.2
52.4 31.6 49.2 33.9 50.9 31.5
7.6 6.9 4.3 4.7 5.8 5.8
63.60 37.50 52.60
1.30 0.85 1.70 0.91 1.59 0.84
Glioblastoma TP-positive (n = 17) TP-negative (n = 10) Total (n = 27)
20.0 9.6 21.3 9.5 20.5 9.4
64.0 31.4 40.0 26.1 56.7 31.4
34.4 28.5 19.4 26.1 22.9 28.1
100.00 100.00 100.00
2.05 0.80 0.98 0.22* 1.69 0.84
All gliomas TP-positive (n = 30) TP-negative (n = 33)
15.6 10.1 10.0 10.4*
56.1 33.2 29.0 26.8**
26.0 27.2 9.6 16.6*
*P < 0.05 **P < 0.01 compared to the TP-positive group
83.30 51.5**
1.81 0.90 1.10 0.80*
150
TP expression and p53 positivities in astrocytic tumors
Fig. 4 Correlation between thymidine phosphorylase and VEGF mRNA expression in human astrocytic tumors evaluated by RT-PCR analysis. There is a signi®cant correlation between TP and VEGF expression (r = 0.75, P < 0.01)
powerful angiogenic factor in gliomas (Takano et al. 1996), as revealed by both immunohistochemical and RT-PCR studies. One of the mechanisms for the correlation between TP and VEGF expression in gliomas could be explained by hypoxic induction of TP as well as VEGF in solid tumors in vitro and in vivo (Griths et al. 1997). These data suggest that TP expression plays an important role in glioma angiogenesis. TP activity was detected in the CSF of hardly any patients with malignant brain tumors including meningeal carcinomatosis (Nakagawa et al. 1998). TP does not possess a signal sequence and is, therefore, not a classical secreted protein (Ishikawa et al. 1989) and a speci®c receptor protein has not been identi®ed. The potential mechanism of action of TP may not require secretion of the protein. Enzyme activity appears to be essential for the stimulatory eect of TP on angiogenesis. The action of this factor within one cell could aect surrounding cells within a tumor as the products/substrates can cross the cell membrane by facilitated diusion/membrane transport (Stevenson et al. 1998). In our study, tumor endothelial cells were predominantly negative and occasionally stained, con®rming a previous study of lung cancer (Giatromanolaki et al. 1997). The precise mechanisms by which TP promotes angiogenesis have been recently demonstrated by changing the relative concentrations of pyrimidine-based compounds and their metabolites in the interstitial ¯uid surrounding endothelial cells (Stevenson et al. 1998). Recently, gliostatin, which was puri®ed from the extracts of neuro®broma, has also been shown to be TP (Asai et al. 1992a). Gliostatin inhibits glial growth, promotes endothelial proliferation, and stimulates neurite outgrowth of embryonic neurons (Asai et al. 1992b). The mechanism of glial growth inhibition by TP/ gliostatin remains to be determined. TP/gliostatin might have alternative eects, becoming angiogenic in particular pathological conditions such as tumors (Fox et al. 1995).
The relationship between TP expression and p53 positivities is not clear. However, wild-type p53 induction in glioma cells down-regulated the endogenous VEGF mRNA level as well as VEGF promoter activity in a dose-dependent manner (Mukhopadhyay et al. 1995). Wild-type p53 induction into glioma cells elicits an inhibitor of endothelial cell migration (Van Meir et al. 1994). These results suggest that wild-type p53 may play a role in suppressing glioma angiogenesis. Kieser et al. (1994) reported that mutated form of the murine p53 induced expression of VEGF mRNA, suggesting that mutated p53 may play a role in stimulating glioma angiogenesis. Also we demonstrated a signi®cant correlation between p53 positivities and vessel density in glioma tissues (data not shown), consistent with the data of mutated p53 stimulating glioma angiogenesis. A similar association of p53 protein overexpression and microvascular density has been observed in head-and-neck squamous cell carcinoma (Gasparini et al. 1993) and colorectal adenocarcinoma (Vermeulen et al. 1996). Our data, showing that TP-positive tumors are signi®cantly positive for p53, supports the hypothesis that TP plays an important role in glioma angiogenesis through p53 overexpression. TP expression and spontaneous apoptosis in astrocytic tumors Our study demonstrated a strong relationship between TP expression and apoptotic indices in astrocytic tumors. TP-positive tumors had higher apoptotic indices than TP-negative tumors. Spontaneous apoptosis is an unfavorable prognostic factor in ovarian cancer (Mattern et al. 1998), endometrial adenocarcinoma (Heatley 1997), and non-small-cell lung carcinoma (Tormanen et al. 1995). There was a positive correlation between apoptotic index and proliferating cell nuclear antigen labeling indices in hepatocellular carcinoma (Hino et al. 1996). With regard to brain tumors, our data showed that the degree of apoptosis was associated with higher MIB-1 positivities and histological malignancy, LGA and AA showed less apoptosis than GBM, con®rming the previous studies (Carroll et al. 1997; Nakagawa et al. 1995; Schier et al. 1995). Accelerated proliferation may cause more frequent DNA duplication and mitotic failures, and cells with damaged DNA die through an apoptotic process (Nakagawa et al. 1995). Otherwise, it can be deduced that many cells in the highturnover state are primed for apoptosis and are, therefore, susceptible to death by this mechanism (Schier et al. 1995). In our study, TP expression was also associated with histological malignancy. In the entire group of astrocytic tumors, TP expression with high apoptotic indices may simply reveal the proliferative potentials and histological malignancy of the tumor.
151
Malignant astrocytic tumors possess hypoxic lesions leading to apoptosis and necrosis. In the hypoxic lesions, TP is induced and may prevent further apoptosis resulting in the progression of the tumors (Kitazono et al. 1998). In glioblastomas, the apoptotic index was higher in TP-positive glioblastomas than in TP-negative glioblastomas, but other tumor variables: MIB-1 positivities, vascular density, p53 positivities, and VEGF expression, were similar in the two types of tumor. These results suggest a link between TP expression and tumor apoptosis in glioblastomas, but the reason for such a correlation is unclear. One explanation is the eect of endogenous angiogenesis inhibitors in glioblastomas. The treatment of solid tumors with angiogenesis inhibitors, including angiostatin and endostatin, resulted in inhibition of tumor growth through enhanced apoptosis (Kirsch et al. 1998; O'Reilly et al. 1997). Also, in cervical cancer, chemotherapy enhanced apoptosis and decreased tumor angiogenesis and TP expression (Ueda et al. 1998). In glioblastomas, not only angiogenic factors, such as VEGF and TP, but also endogenous angiogenesis inhibitors, such as thrombospondin-1, interferon-inducible protein 10 (IP-10) and brain-speci®c angiogenesis inhibitor-1 (BAI-1) are up-regulated (unpublished data). The antitumor activity of thrombospondin-1 may result from an increased sensitivity to apoptosis in endothelial cells adjacent to a provisional matrix during formation of vascular beds in tumors expressing TSP-1 (Guo et al. 1997). These endogenous angiogenesis inhibitors may aect the ratio of spontaneous apoptosis in glioblastomas, where angiogenic activity through VEGF and TP in glioblastomas could overcome anti-angiogenic activity. Further studies concerning the relationship between angiogenic balance and apoptosis in glioblastomas will be needed.
Summary TP was found to be a potent angiogenic factor closely associated with cell proliferation and tumor apoptosis in human astrocytic tumors. TP as well as VEGF expression on highly vascularized astrocytic tumors could be an eective target for future therapeutic endeavors in antiangiogenesis. Acknowledgements We are grateful to Nippon Roche Research Center for the gift of anti-thymidine phosphorylase antibody and for assistance with ELISA. We also gratefully acknowledge the critical reading of the manuscript by Yutaka Tanaka (Nippon Roche Research Center) and the excellent technical assistance of Yoshiko Tsukada, Kazuko Morita, and Makiko Miyagawa. This study was supported by Grant-in Aid 10671282 from the Japanese Ministry of Education, Science, and Culture (to S.T.).
References Ansari B, Coates PJ, Greenstein BD, Hall PA (1993) In situ endlabeling detects DNA strand breaks in apoptosis and other physiological and pathological states. J Pathol 170:1±8
Asai K, Hirano T, Kaneko S, Moriyama A, Nakanishi K, Isobe I, Eksioglu YZ, Kato T (1992a) A novel glial growth inhibitory factor, gliostatin, derived from neuro®broma. J Neurochem 59:307±317 Asai K, Nakanishi K, Isobe I, Eksioglu YZ, Hirano A, Hama K, Miyamoto T, Kato T (1992b) Neurotrophic action of gliostatin on cortical neurons. J Biol Chem 267:20311±20316 Carroll RS, Zhang J, Chauncey BW, Chantziara K, Frosch MP, Black PM (1997) Apoptosis in astrocytic neoplasms. Acta Neurochir 139:845±850 Fox SB, Moghaddam A, Westwood M, Turley H, Bicknell R, Gatter KC, Harris AL (1995) Platelet-derived endothelial cell growth factor/thymidine phosphorylase expression in normal tissues: an immunohistochemical study. J Pathol 176:183±190 Gasparini G, Weidner N, Maluta S, Pozza F, Boracchi P, Mezzetti M, Testolin A, Bevilacqua P (1993) Intratumoral microvessel density and p53 protein: correlation with metastasis in headand-neck squamous-cell carcinoma. Int J Cancer 55:739±744 Giatromanolaki A, Koukourakis MI, Comley M, Kaklamanist L, Turley H, O'Byrne K, Harris AL, Gatter KC (1997) Plateletderived endothelial cell growth factor (thymidine phosphorylase) expression in lung cancer. J Pathol 181:196±199 Griths L, Dachs GU, Bicknell R, Harris AL, Stratford IJ (1997) The in¯uence of oxygen tension and pH on the expression of platelet-derived endothelial cell growth factor/thymidine phosphorylase in human breast tumor cells grown in vitro and in vivo. Cancer Res 57:570±572 Guo N, Krutzsch HC, Inman JK, Roberts DD (1997) Thrombospondin 1 and type 1 repeat peptides of thrombospondin 1 speci®cally induce apoptosis of endothelial cells. Cancer Res 57:1735±1742 Haraguchi M, Miyadera K, Uemura K, Sumizawa T, Furukawa T, Yamada K, Akiyama S (1994) Angiogenic activity of enzymes. Nature 368:198±199 Heatley MK (1997) A high apoptotic index occurs in subtypes of endometrial adenocarcinoma associated with a poor prognosis. Pathology 29:272±275 Hino N, Higashi T, Nouso K, Nakatsukasa H, Tsuji T (1996) Apoptosis and proliferation of human hepatocellular carcinoma. Liver 16:123±129 Imazono Y, Takebayashi Y, Nishiyama K, Akiba S, Miyadera K, Yamada Y, Akiyama S, Ohi Y (1997) Correlation between thymidine phosphorylase expression and prognosis in human renal cell carcinoma. J Clin Oncol 15:2570±2578 Ishikawa F, Miyazono K, Hellman U, Drexler H, Wernstedt C, Hagiwara K, Usuki K, Takaku F, Risau W, Heldin C (1989) Identi®cation of angiogenic activity and the cloning and expression of platelet-derived endothelial cell growth factor. Nature 338:557±562 Kieser A, Weich HA, Brandner G, Marme D, Kolch W (1994) Mutant p53 potentiates protein kinase C induction of vascular endothelial growth factor expression. Oncogene 9:963±969 Kirsch M, Strasser J, Allende R, Bello L, Zhang J, Black PM (1998) Angiostatin suppresses malignant glioma growth in vivo. Cancer Res 58:4654±4659 Kitazono M, Takebayashi Y, Ishitsuka K, Takao S, Tani A, Furukawa T, Miyadera K, Yamada Y, Aikou T, Akiyama S (1998) Prevention of hypoxia-induced apoptosis by the angiogenic factor thymidine phosphorylase. Biochem Biophys Res Commun 253:797±803 Koukourakis MI, Giatromanolaki A, O'Byrne KJ, Comley M, Whitehouse RM, Talbot DC, Gatter KC, Harris AL (1997) Platelet-derived endothelial cell growth facto expressions correlates with tumour angiogenesis and prognosis in non-smallcell lung cancer. Br J Cancer 75:477±481 Kubota Y, Miura T, Moriyama M, Noguchi S, Matsuzaki J, Takebayashi S, Hosaka M (1997) Thymidine phosphorylase activity in human bladder cancer: dierence between super®cial and invasive cancer. Clin Cancer Res 3:973±976 Mattern J, Stammler G, Koomagi R, Wallwiener D, Kaufmann M, Volm M (1998) Spontaneous apoptosis in ovarian cancer: an unfavorable prognostic factor. Int J Oncol 12:351±354
152 Moghaddam A, Bicknell R (1992) Expression of platelet-derived endothelial cell growth factor in Escherichia coli and conformation of its thymidine phosphorylase activity. Biochemistry 31:12 141±12 146 Moghaddam A, Zhang H, Fan TD, Hu D, Lees VC, Turley H, Fox SB, Gatter KC, Harris AL, Bicknell R (1995) Thymidine phosphorylase is angiogenic and promotes tumor growth. Proc Natl Acad Sci USA 92:998±1002 Mukhopadhyay D, Tsiokas L, Sukhatme VP (1995) Wild-type p53 and v-Src exert opposing in¯uences on human vascular endothelial growth factor gene expression. Cancer Res 55:6161±6165 Nakagawa S, Shiraishi T, Kihara S, Tabuchi K (1995) Detection of DNA strand breaks associated with apoptosis in human brain tumors. Virchows Arch 427:175±179 Nakagawa H, Yamada M, Fukushima M, Shimizu K, Ikenaka K (1998) In vitro study on intrathecal application of 5-¯uoro-2¢deoxyuridine (FdUrd) for meningeal dissemination of malignant tumor. No Shinkei Geka 26:787±794 Nakayama Y, Sueishi K, Fukushima T, Oka K, Tomonaga M (1994) Localization of platelet-derived endothelial cell growth factor in human glioblastoma and meningioma. Noshuyo Byori 11:187±191 Ng SY, Gunning P, Eddy R, Ponte P, Leavitt J, Shows T, Kedes L (1985) Evolution of the functional human b-actin gene and its multipseudogene family: conservation of non-coding regions and chromosomal dispersion of pseudogenes. Mol Cell Biol 5:2720±2732 Nishida M, Hino A, Mori K, Matsumoto T, Yoshikubo TT, Ishitsuka H (1996) Preparation of anti-human thymidine phosphorylase monoclonal antibodies useful for detecting the enzyme levels in tumor tissues. Biol Pharm Bull 19:1407± 1411 O'Brien TS, Fox SB, Dickinson AJ, Turley H, Westwood M, Moghaddam A, Gatter KC, Bicknell R, Harris AL (1996) Expression of the angiogenic factor thymidine phosphorylase/ platelet-derived endothelial cell growth factor in primary bladder cancers. Cancer Res 56:4799±4804 O'Reilly MS, Boehm T, Shing Y, Fukai N, Vasios G, Lane WS, Flynn E, Birkhead JR, Olsen BR, Folkman J (1997) Endostatin: an endogenous inhibitor of angiogenesis and tumor growth. Cell 88:277±285 Relf M, LeJeune S, Scott PA, Fox S, Smith K, Leek R, Moghaddam A, Whitehouse R, Bicknell R, Harris AL (1997) Expression of the angiogenic factors vascular endothelial cell growth factor, acidic and basic ®broblast growth factor, tumor growth factor beta-1, platelet-derived endothelial cell growth factor, placenta growth factor, and pleiotrophin in human primary breast cancer and its relation to angiogenesis. Cancer Res 57:963±969 Schier D, Cavalla P, Migheli A, Chio A, Giordana MT, Marino S, Attanasio A (1995) Apoptosis and cell proliferation in human neuroepithelial tumors. Neurosci Lett 195:81±84
Sotos GA, Grogan L, Allegra CJ (1994) Preclinical and clinical aspects of biomodulation of 5-¯uorouracil. Cancer Treat Rev 20:11±49 Stevenson DP, Milligan SR, Collins WP (1998) Eects of plateletderived endothelial cell growth factor/thymidine phosphorylase, substrate, and products in a three-dimensional model of angiogenesis. Am J Pathol 152:1641±1646 Takano S, Yoshii Y, Kondo S, Maruno T, Shirai S, Nose T (1996) Concentration of vascular endothelial growth factor in the serum and tumor tissue of brain tumor patients. Cancer Res 56:2185±2190 Takano S, Wadhwa R, Yoshii Y, Nose T, Kaul SC, Mitsui Y (1997) Elevated levels of mortalin expression in human brain tumors. Exp Cell Res 237:38±45 Takebeyashi Y, Akiyama S, Akiba S, Yamada K, Miyadera K, Sumizawa T, Yamada Y, Murata F, Aikou T (1996) Clinicopathologic and prognostic signi®cance of an angiogenic factor, thymidine phosphorylase, in human colorectal carcinoma. J Natl Cancer Inst 88:1110±1117 Tanigawa N, Amaya H, Matsumura M, Katoh Y, Kitaoka A, Aotake T, Shimomatsuya T, Rosenwasser OA, Iki M (1996) Tumor angiogenesis and expression of thymidine phosphorylase/platelet derived endothelial cell growth factor in human gastric carcinoma. Cancer Lett 108:281±290 Toi M, Hoshina S, Taniguchi T, Yamamoto Y, Isahitsuka H, Tominaga T (1995) Expression of platelet-derived endothelial cell growth factor/thymidine phosphorylase in human breast cancer. Int J Cancer 64:79±82 Tormanen U, Eerola AK, Rainio P, Vahakangas K, Soini Y, Sormunen R, Bloigu R, Lehto VP, Paakko P (1995) Enhanced apoptosis predicts shortened survival in non-small cell lung carcinoma. Cancer Res 55:5595±5602 Ueda M, Ueki K, Kumagai K, Terai Y, Okamoto Y, Ueki M, Otsuki Y (1998) Apoptosis and tumor angiogenesis in cervical cancer after preoperative chemotherapy. Cancer Res 58:2343± 2346 Usuki K, Gonez LJ, Wernstedt C, Moren A, Miyazono K, Claesson-Welsh L, Heldin CH (1994) Structural properties of 3.0 kb and 3.2 kb transcripts encoding platelet-derived endothelial cell growth factor/thymidine phosphorylase in A431 cells. Biochim Biophys Acta 1222:411±414 Van Meir EG, Polverini PJ, Chazin VR, Huang HJS, Tribolet N de, Cavenee WK (1994) Release of an inhibitor of angiogenesis upon induction of wild type p53 expression in glioblastoma cells. Nat Genet 8:171±176 Vermeulen PB, Roland L, Mertens V, Marck EV, De Bruijin EA, Van Oosterom AT, Dirix LY (1996) Correlation of intratumoral microvessel density and p53 protein overexpression in human colorectal adenocarcinoma. Microvasc Res 51:164±174 Weindel K, Marme D, Weich HA (1992) AIDS-associated Kaposi's sarcoma cells in culture express vascular endothelial growth factor. Biochem Biophys Res Commun 183:1167±1174
