Metastasis suppressor CC3 inhibits angiogenic properties of ... - Nature

ã

Oncogene (2001) 20, 270 ± 275 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

SHORT REPORT

Metastasis suppressor CC3 inhibits angiogenic properties of tumor cells in vitro Roisin NicAmhlaoibh1 and Emma Shtivelman*,1 1

Cancer Research Institute, University of California San Francisco, 2340 Sutter Street, San Francisco, California, CA 94143-0128, USA

Resistance to apoptosis and ability to promote angiogenesis are integral features of the metastatic phenotype. Human gene CC3 is a metastasis suppressor for variant small cell lung carcinoma and a mouse melanoma in vivo. We have shown previously that metastasis-suppressing function of CC3 might be due at least in part to the ability of CC3 protein to predispose tumor cells to apoptosis. Here we demonstrate that CC3 has a previously unidenti®ed eect on the ability of tumor cells to induce angiogenesis in vitro. Expression of CC3 in three dierent tumor cell lines signi®cantly diminished their angiogenic character as manifested in the in vitro proliferation and migration assays with endothelial cells of both macro- and microvascular origin. Expression of CC3 induced changes in RNA levels of several angiogenic modulators consistent with the overall reduction in angiogenic properties. These results indicate that expression of CC3 has a dual eect on phenotype of tumor cells ultimately inhibiting their metastatic potential. Oncogene (2001) 20, 270 ± 275. Keywords: angiogenesis; metastasis; CC3

The process of angiogenesis or sprouting of new capillaries from existing vessels is essential to tumor growth and metastasis. To stimulate angiogenesis tumor cells secrete a number of angiogenic factors such as ®broblast growth factors (aFGF and bFGF) and vascular endothelial cell growth factor (VEGF). Tumors also generate antiangiogenic factors among them thrombospondin (Good et al., 1990), angiostatin (O'Reilly et al., 1994), endostatin (O'Reilly et al., 1997) and others. The net balance between positive and negative regulators determines whether a tumor develops a new blood supply and means of survival and growth. The ability to induce and support angiogenesis is of particular importance for growth of micrometastatic lesions (Holmgren et al., 1995). Human gene CC3 was identi®ed as a novel metastasis suppressor gene of variant small cell lung

*Correspondence: E Shtivelman Received 10 July 2000; revised 25 October 2000; accepted 1 November 2000

carcinoma or SCLC (Shtivelman, 1997). SCLC is a highly metastatic lung malignancy with neuroendocrine features (Dearing et al., 1990). A very aggressive subtype known as variant SCLC (v-SCLC) has an even worse prognosis than classic SCLC (c-SCLC). In vivo studies using SCID-hu mice metastasis model showed that v-SCLC have a signi®cantly higher metastatic potential that c-SCLC (Shtivelman and Namikawa, 1995). Expression of CC3 is absent in highly metastatic v-SCLC cells but found in the less metastatic c-SCLC tumor lines, in other tumors and in normal cells (Shtivelman, 1997). Introduction of CC3 expression suppresses metastasis of v-SCLC cells in SCID-hu mice (Shtivelman, 1997) and of mouse melanoma B16 in a syngeneic metastasis model (Liu et al., 1999). Analysis of the consequences of constitutive expression of CC3 in tumor cells showed that CC3 sensitizes v-SCLC cells to apoptosis induced by a variety of death signals (Shtivelman, 1997, Whitman et al., 2000). Loss of CC3 may increase survival of metastatic cells in circulation or within a micrometastatic lesion by conferring resistance to apoptosis induced by altered microenvironment. CC3 encodes a relatively small protein of 30 kD that is highly conserved in evolution (Shtivelman, 1997). Amino acid sequence of CC3 contains a weak but signi®cant homology to a large family of enzymes known as short chain dehydrogenases/reductases though no enzymatic function was demonstrated for CC3 protein (Baker, 1999). CC3 was identi®ed independently as a cellular protein capable of binding to the HIV Tat protein and contributing to the transcriptional eects of the latter (Xiao et al., 1998). In addition, deregulated expression of CC3 in v-SCLC cells was shown to lead to changes in expression levels of some cellular genes implying that CC3 might play a role in transcriptional regulation (Xiao et al., 2000). In this study, we set to examine the potential eect of CC3 on angiogenic properties of tumor cells in vitro as a possible factor in the metastasis-suppressing ability of CC3. First, we have examined the in vitro angiogenic character of SCLC lines whose metastatic potential correlates inversely with expression of CC3. To determine if the dierence in metastatic potential of v-SCLC and c-SCLC lines described earlier (Shtivelman and Namikawa, 1995) might correlate with the dierences in their angiogenic properties, we tested conditioned media (CM) from both types of SCLC

CC3-induced inhibition of angiogenesis R NicAmhlaoibh and E Shtivelman

cells in proliferation assays with human umbilical vein endothelial cells (HUVEC). Two cell lines N417 and H82 are representative of the highly metastatic v-SCLC and the other two, H69 and H146, are of less metastatic c-SCLC type (Shtivelman and Namikawa, 1995). We found that CM from v-SCLC lines N417 and H82 stimulated HUVEC proliferation rate above that observed with control medium. [3H]thymidine incorporation in HUVEC cells was increased by 45 and 16% respectively (Figure 1). However, HUVEC cells treated with CM collected from c-SCLC lines failed to achieve proliferation levels observed in control cultures. [3H]thymidine incorporation was reduced by 48% for H69 cells and 28% for H146 cells (Figure 1). The observed dierences are unlikely to stem from the dierent proliferation rates among the cell lines used, because lines N417, H82 and H69 have similar doubling times with only H146 cells showing a signi®cantly slower proliferation rate (by about 30%). These results on dierent angiogenic ability in vitro of v-SCLC versus c-SCLC lines are suggestive but they do not provide conclusive evidence for a relationship between angiogenic activity in vitro, metastasis in vivo of SCLC cells and expression of metastasis suppressor CC3. Metastatic potential of SCLC lines inversely correlates with expression of CC3 that is found in the cSCLC lines H69 and H146 but not in N417 or H82; expression of ectopic CC3 suppresses metastasis of

N417 cells (Shtivelman, 1997). We have therefore addressed the possibility that CC3 downmodulates angiogenic properties of N417 cells as a possible mechanism contributing to metastasis suppression. We have also decided to examine if expression of exogenous CC3 in CC3-negative or low metastatic cells other than v-SCLC will result in suppression of their angiogenic properties. CC3 expression was stably introduced into three tumor cell lines with undetectable levels of endogenous CC3 RNA and protein: v-SCLC N417, neuroblastoma SKNSH and glioblastoma U373; as well as melanoma C32 which has very low endogenous CC3 levels (Figure 2). Cells were transfected with expression constructs for CC3 with or without epitope tags and expression was examined by RT ± PCR (Figure 2a) and by Western blot using both the antibody to the epitope tag where appropriate (not shown) or a polyclonal antibody raised against CC3 (Figure 2b). Cell clones with high levels of exogenous CC3 protein (shown in Figure 2) were chosen and used in experiments designed to evaluate potential eect of CC3 on secretion of angiogenic modulators. We found that in three out of four cell lines expression of CC3 resulted in inhibition of HUVEC proliferation compared to that seen with CM of control clones (Figure 3a). A signi®cant inhibitory eect was seen with N417 (2.7+0.9-fold), SKNSH (2.9+1.4-fold) and C32 (2.4+1.6-fold) but not with U373 (1.1+0.1-fold inhibition). A similar inhibitory

Figure 1 Inhibition of endothelial cell proliferation by conditioned media from CC3-expressing c-SCLC cell lines. HUVEC cells were purchased from Clonetics Corporation, maintained in fully supplemented EGM-2 medium and used before or at sixth passage. To produce conditioned medium from tumor lines, cells were seeded at a density of 16105 cells per ml of RPMI-1640 supplemented with 10% FBS. Cells were allowed to proliferate for 2 days. The media were collected, aliquoted and stored at 7808C. HUVEC were seeded at a density of 1000 cells/well in a 96-well plate. Cells were allowed to adhere for 4 ± 6 h at which time the media was removed and 100 ml of a 1 : 1 mixture of EBM basal medium (Clonetics) and conditioned media (or control RPMI1640 with 10% FBS) was added to each well. The cells were left to proliferate for 4 days and 0.25 mCi of [3H]thymidine was added for the ®nal 24 h. Cell proliferation (DNA synthesis) was measured as [3H]thymidine incorporation in c.p.m. per well. Each reading was an average of six replica wells. Data are presented as % of [3H]thymidine incorporation relative to control (RPMI-1640 with 10% serum instead of conditioned medium). Each bar represents the average of at least three independent experiments

Figure 2 Expression of CC3 in transfected cell lines. (a) Analysis of CC3 RNA in transfected cell lines by RT ± PCR. CC3-speci®c primers corresponded to the sequences around initiation and stop codons. Ampli®cation product is 726 bp in length and represents the entire open reading frame of CC3 RNA. Endogenous CC3 RNA is detectable only in C32 cells. The lower band corresponds to ampli®ed b-actin RNA sequences used as an internal control for PCR. (b) Western blot analysis performed with total protein extracts prepared with cell lines indicated. Polyclonal serum raised against bacterially expressed CC3 protein was used. Dierent sizes of exogenous CC3 proteins are due to the fact that CC3 introduced into SKNSH and C32 cells had three HA epitopes at the C terminus, U373cc3 expresses exogenous CC3 with one HA epitope, while N417cc3 cells express untagged exogenous CC3 which therefore has same molecular weight as endogenous CC3 weakly expressed in C32 lines. No endogenous CC3 protein could be detected in N417, SKNSH or U373 cells even after prolonged exposure

271

Oncogene


272

20 ng/ml VEGF added to control medium (Figure 4). However, CM of N417cc3 cells failed to stimulate migration of endothelial cells above that seen with the control medium. Although CM from SKNSH cells did not signi®cantly stimulate HUVEC migration, conditioned media from the CC3-expressing clone was found to inhibit migration almost threefold. The speci®city of the inhibitory eect of CM of N417cc3 cells towards endothelial cells was examined with primary and established cells of dierent origins. Cells of endothelial and non-endothelial origin (Table 1) were tested for proliferation in the presence of CM of N417neo and N417cc3 cells. We found that the inhibitory eect of N417cc3 CM is observed with human lung microvascular endothelial cells (HMVECL) and with vascular smooth muscle cells (PASMC), a second cellular component of blood vessels. Inhibitory eect was less pronounced with normal human dermal ®broblasts (NHDF) and not signi®cant with immortalized NHDF line (NHDF-hT), immortalized rodent

Figure 3 Inhibition of endothelial cell proliferation by conditioned media from CC3-expressing tumor cells. (a) HUVEC cells were grown in conditioned media collected from N417(N), SKNSH (SK), C32 (C) and U373 (U) single cell clones stably transfected with empty vector (neo) or expressing ectopic CC3 protein (cc3). Proliferation assays were conducted and are presented as described in legend to Figure 1. Statistical analysis of results show following P values for observed dierences in proliferation rates: 0.0039 for Nneo versus Ncc3 (n=9); 0.014 for SKneo versus SKcc3 (n=8) and 0.169 for Cneo versus Ccc3 (n=6). (b) For cell counts, HUVEC cells were seeded at a density of 30 000 cells per well in a six-well plate, allowed to proliferate in control or conditioned medium for 2 or 4 days, collected by trypsinization and counted. Each bar represents the average of two independent experiments

eect on HUVEC proliferation was seen with CM collected from other clonal populations of N417cc3 and SKNSHcc3 (not shown). These results were further con®rmed by performing cell counts on CMtreated HUVEC (Figure 3b). However, no decrease in cell viability was observed for cells grown in any of the CM, indicating that decreased cell number is not due to cell death (data not shown). Endothelial cell migration is essential to the development of a new network of blood vessels to support tumor growth. To determine if CC3 expression leads to changes in ability of tumor cells to stimulate endothelial cell migration, cell migration/invasion assays were performed. HUVEC were allowed to migrate through a collagen interface towards CM collected from N417 and SKNSH clones. CM of N417 was found to be highly chemotactic for HUVEC stimulating their migration about twofold over control and to a greater extent than either 5 ng/ml bFGF or Oncogene

Figure 4 Inhibition of HUVEC migration by conditioned media from CC3-expressing tumor cells. Modi®ed Boyden chambers with 10 mm thick polycarbonate membranes and 8 mm pores were used. The upper side of the membranes was coated with 10 mg/ml collagen type I in PBS for 2 h at 378C. The membranes were rinsed twice with PBS and placed into chambers containing 500 ml of conditioned media, control media (RPMI-1640 with 10% FBS) or PBS. Subcon¯uent HUVEC cells were trypsinized and resuspended in basal medium with 0.5% BSA at a density of 56105 cells/ml. 100 ml of HUVEC cell suspension was added to the top of each migration chamber. Cells were allowed to migrate to the underside of the top chamber overnight at 378C in 5% CO2. Cells on the upper membrane were removed with a cotton swab, and the migratory cells attached to the underside of the membrane were ®xed, stained and counted under microscope using a 206objective lens. Cell migration was measured as the number of cells that migrated overnight towards conditioned media of indicated tumor lines, fresh RPMI-1640/10% serum media (control), or control media supplemented with 5 ng/ml bFGF (control+bFGF) or 20 ng/ml VEGF (control+VEGF). No migration towards PBS was observed in any of the experiments. Each bar represents the average of four or ®ve independent experiments. Statistical analysis of results revealed following P values for observed dierences in migration rates: 0.001 for Nneo versus Ncc3 (n=5) and 0.04 for SKneo versus SKcc3 (n=4)


273

Table 1 Inhibitory eect of conditioned media from CC3-expressing N417 cells on endothelial and non-endothelial cell proliferation Cell type

Cell line

Endothelial

HUVEC HMVEC-L PASMC NHDF NHDF-hT Rat1a NIH3T3 MDA MB 435 SKNSH C32

Vascular smooth muscle Normal fibroblasts Immortalized fibroblasts Tumor cell lines

Fold inhibition 2.7+0.9a 2.9+1.6b 2.6+1.1c 1.9+0.6* 1.5+0.1* 1.1+0.1 1.3+0.1 1.1+0.1 1.0+0.1 1.0+0.1

Conditioned media from N417neo and N417cc3 cells was applied to a variety of endothelial and non-endothelial cells in a 4 day proliferation assay. All primary cells were purchased from Clonetics and maintained in recommended medium (HMVEC-L in EGM2MV, HUVEC in EGM-2, PASMC in SmGM-2 and NHDF in FGM-2 bullet kits). The immortalized NHDF (NHDF-hT) cell line, a gift of Dr E Venetsanakos, was maintained as NHDF. All primary human cells were used within three or four passages of receipt from supplier. Fold inhibition was determined as the fold dierence of [3H]thymidine incorporation between cells grown in N417 conditioned media and N417cc3 conditioned media. Statistical analysis of dierences shows aP-value of 0.0039 (n=9); bP-value of 0.0313 (n=6), cP-value of 0.0156 (n=7); *Accurate analyses could not be carried out as n56

®broblasts Rat1a or NIH3T3 or human tumor cell lines (MDA-MB-435, SKNSH, C32). The inhibitory eect of CM from CC3 expressing cells may be due to a reduction of a stimulatory component secreted by CC3-negative parental cells rather than secretion of an active inhibitor. To address this issue endothelial cell proliferation assays were carried out in the presence of known stimulators of proliferation bFGF and VEGF. Both factors increased absolute levels of [3H]thymidine incorporation when added to CM. However, the inhibitory eect of the N417cc3 CM relative to N417neo CM remained intact (Figure 5a,b). This indicates that an active inhibitor of proliferation is likely to be secreted by CC3 expressing cells. We have examined expression of several known angiogenic modulators in clones of N417, SKNSH and C32 cells by RT ± PCR analyses. Expression of RNAs for several known inhibitors of angiogenesis was analysed by RT ± PCR. We have found that the following gene products are either not expressed in cell lines examined or their expression is not aected by expression of CC3: thrombospondin (Good et al., 1990), thrombospondin-like METH-1 and 2 (VaÂzquez et al., 1999), PEDF (Dawson et al., 1999), VEGI (Zhai et al., 1999), MIG (Addison et al., 2000), IP10 (Strieter et al., 1995), and platelet factor 4 (Gentilini et al., 1999). Expression of angiogenic stimulators bFGF, VEGF and TGFb1 was detected in all cell lines but was not aected by exogenous CC3 (not shown). Several factors listed below were found to be aected by CC3 in at least one of the pair of cell lines and those were further examined by Northern blot analysis (Figure 6a). Expression of angiogenic stimulator angiopoietin 1 (Koblizek et al., 1998) was markedly

Figure 5 Inhibitory eect of conditioned media from CC3expressing N417 cells on endothelial cell proliferation is maintained in presence of bFGF or VEGF. (a) Conditioned media collected from N417neo and N417cc3 cells was applied to HUVEC cells in the presence or absence of 5 ng/ml bFGF in a 4 day proliferation assay. (b) Conditioned media collected from N417 and Ncc3 cells was applied to HMVEC-L cells (Clonetics Corporation) in the presence or absence of 20 ng/ml VEGF in a 4 day proliferation assay. Proliferation was measured as [3H]thymidine incorporation in c.p.m. per well

reduced in C32cc3 cells. RNA for angiopoietin 2, an antagonist of angiopoietin 1 (Maisonpierre et al., 1997; Witzenbichler et al., 1998) was upregulated by CC3 in SKNSH and C32 cells. Low level of RNA for angiogenic inhibitor BAI1 (Nishimori et al., 1997) found in N417 cells was elevated in N417cc3 cells. Expression of RNA for inhibitor of angiogenesis SPARC (Kupprion et al., 1998) was somewhat elevated in SKNSHcc3 cells. All these changes are consistent with the reduced angiogenic character of cells forced to express CC3. VEGF mRNA was also examined by Northern blot analysis in all cell lines though RT ± PCR did not indicate any changes in its levels depending on CC3. We found that VEGF levels are somewhat reduced in N417 and SKNSH cells in presence of CC3 but are slightly increased in C32cc3 cells. The changes were apparently not signi®cant enough to be detected by RT ± PCR. The Northern blot results were veri®ed by Western blot analysis (Figure 6b) which detected very slight changes in expression of VEGF protein consistent with changes of VEGF RNA. We suggest that small changes in levels of VEGF induced by expression of CC3 (including a small upregulation in C32cc3 cells) might be too insigni®cant to play a key role in its eects on angiogenic properties. This is corroborated Oncogene


274

Figure 6 Eect of CC3 on RNA levels of angiogenic modulators. (a) Northern blot analysis of expression of angiogenic modulators in control (neo) or CC3 expressing clones (cc3). Two identical blots prepared simultaneously and containing 2 mg of polyadenylated RNA per lane were used; SPARC probe was hybridized to a blot that contained 5 mg total RNA. All blots were reprobed with an unrelated probe (MCM5) to control for loading of RNA. (b) Western blot analysis of VEGF protein was performed with a rabbit polyclonal serum (Santa Cruz Biotechnology)

by the data described above that addition of high levels of VEGF or bFGF did not abolish inhibitory activity contained in CM of N417cc3 cells (Figure 5). The metastasis suppressing action of CC3 has been ascribed in part to the ability of this protein to induce sensitivity to apoptotic stimuli in originally nonresponsive tumor cells (Shtivelman, 1997). In this study, we addressed the possibility that CC3 might adversely aect another important property of metastatic cells, i.e. their ability to stimulate angiogenesis. We show indeed that forced expression of CC3 in tumor cells has an inhibitory eect on their angiogenic characteristics in vitro. We propose that CC3 may have a dual eect on tumor cells promoting their apoptosis and inhibiting angiogenic properties thus preventing the establishment of metastases. We show here that conditioned media collected from CC3-expressing tumor cells inhibits proliferation of macrovascular and microvascular endothelial cells, vascular smooth muscle cells and to a lesser extent normal ®broblasts. Migration of HUVEC is also diminished or inhibited by CM of CC3-expressing cells. This eect is not due to secretion of CC3 protein itself because repeated attempts to detect CC3 polypeptides in conditioned medium have not produced positive results in eect ruling out the possibility of CC3 being secreted (not shown). The inhibitory eects described here were maintained in proliferation assays carried out in the presence of bFGF or VEGF indicating that CC3Oncogene

expressing cells secrete an active inhibitor and not just lower levels of angiogenic stimulants. Indeed, we have found that mRNA levels of both angiogenic stimulants and inhibitors are aected by expression of CC3. The inhibitory protein BAI1 is upregulated in v-SCLC N417. In melanoma C32 expression of CC3 leads to downmodulation of RNA for angiopoietin 1 and at the same time upregulation of inhibitory angiopoietin 2. Levels of RNA for angiopoietin 2 and SPARC are elevated in neuroblastoma SKNSH expressing CC3 protein. It is likely that other angiogenic modulators might be aected by expression of CC3. It is interesting that CC3 is able to suppress angiogenic characteristics of three cell lines of dierent origins that were chosen on the basis of undetectable or low expression of endogenous CC3. Introduction of CC3 into these lines also results in increased susceptibility to apoptosis (Whitman et al., 2000 and unpublished results). CM from CC3 expressing cells inhibits proliferation of endothelium, vascular smooth muscle cells and to a lesser extent primary ®broblasts but not immortalized ®broblasts or transformed cells. Unlike speci®c inhibitors of endothelial cell proliferation such as endostatin (O'Reilly et al., 1997) and angiostatin (O'Reilly et al., 1994), other modulators (aFGF, bFGF, angiogenin, thrombospondin, PDECGF, prolactin, TGF-b) aect multiple cell types. It is apparent that CC3 expression aects levels of expression of multiple genes whose products have biological activity towards endothelium but also towards other cell lineages. Some inhibitors of angiogenesis are produced following post-translational modi®cations (Hanahan and Folkman, 1996). Nevertheless we chose to examine eects of CC3 on mRNA levels of angiogenic modulators due to the existing evidence that CC3 acts as a transcriptional co-factor. Xiao et al., (1998) showed that CC3 or TIP30 acts as a cofactor in HIV-1 Tat-activated transcription. More recently it was reported that ectopic expression of CC3/TIP30 in v-SCLC cells induced expression of two pro-apoptotic genes and some other transcripts but also caused a signi®cant downregulation of expression of other genes (Xiao et al., 2000). We have described here signi®cant changes in RNA levels of several angiogenic modulators induced by CC3. All cell lines examined here exhibited changes in expression levels of at least two genes whose products are active towards endothelial cells. However, not a single product was aected in all three lines examined. We suggest that the eects of CC3 on transcription are likely to be indirect and largely dependent on activity of other authentic transcription factor(s). We show that inhibitor of angiogenesis BAI1 identi®ed originally as a transcriptional target of p53 (Nishimori et al., 1997) is induced by CC3 in p53negative N417 cells. Another p53-responsive gene, PIG12, was shown to be induced in N417 cells (Xiao et al., 2000). Our ®nding adds further to the speculation that p53 might act as an upstream


regulator in a CC3-mediated signaling pathway (Xiao et al., 2000). In conclusion, we have shown that ectopic CC3 expression in three tumor cell lines leads to diminution of their ability to induce proliferation and migration of vascular cells in vitro. It is ®rmly accepted that in the absence of angiogenesis, apoptosis suppresses uncontrolled proliferation of micro-metastases (reviewed in Hanahan and Folkman, 1996). The in vitro evidence support a model in which CC3 acts to inhibit angiogenesis within metastatic lesions as well as to promote apoptosis of tumor cells. These two eects would be synergistic in suppressing development of metastases. Further investigation will be required to

determine the exact mechanisms whereby metastasis suppressor CC3 executes its role as an inducer of tumor cell apoptosis and inhibitor of angiogenesis.

275

Acknowledgments We are grateful to Dr Eleni Venetsanakos for the gift of immortalized human ®broblasts, to Dr Nishimori for the BAI1 cDNA, to Josh Sussman for statistical analysis and to Drs Karen Smith-McCune and Robert J. Debs for the critical reading of the manuscript. This work was supported by funds from NIH grant RO1 CA71422 and an award from the Stewart Foundation, UCSF.

References Addison CL, Arenberg DA, Morris SB, Xue YY, Burdick MD, Mulligan MS, Iannettoni MD and Strieter RM. (2000). Hum. Gene. Ther., 11, 247 ± 261. Baker ME. (1999). Curr Biol., 9, R471. Dawson DW, Volpert OV, Gillis P, Crawford SE, Xu H, Benedict W and Bouck NP. (1999). Science, 285, 245 ± 248. Dearing MP, Steinberg SM, Phelps R, Anderson MJ, Mulshine JL, Ihde DC and Johnson BE. (1990). J. Clin. Oncol., 8, 1042 ± 1049. Gentilini G, Kirschbaum NE, Augustine JA, Aster RH and Visentin GP. (1999). Blood, 93, 25 ± 33. Good DJ, Polverini PJ, Rastinejad F, Le Beau MM, Lemons RS, Frazier WA and Bouck NP. (1990). Proc. Natl. Acad. Sci. USA, 87, 6624 ± 6628. Hanahan D and Folkman J. (1996). Cell, 86, 353 ± 364. Holmgren L, O'Reilly MS and Folkman J. (1995). Nat. Med., 1, 149 ± 153. Koblizek TI, Weiss C, Yancopoulos GD, Deutsch U and Risau W. (1998). Curr. Biol., 8, 529 ± 532. Kupprion C, Motamed K and Sage EH. (1998). J. Biol. Chem., 273, 29635 ± 29640. Liu Y, Thor A, Shtivelman E, Cao Y, Guanghuan T, Heath TD and Debs RJ. (1999). J. Biol. Chem., 274, 13338 ± 13344. Maisonpierre PC, Suri C, Jones PF, Bartunkova S, Wiegand SJ, Radziejewski C, Compton D, McClain J, Aldrich TH, Papadopoulos N, Daly TJ, Davis S, Sato TN and Yancopoulos GD. (1997). Science, 277, 55 ± 60. Nishimori H, Shiratsuchi T, Urano T, Kimura Y, Kiyono K, Tatsumi K, Yoshida S, Ono M, Kuwano M, Nakamura Y and Tokino T. (1997). Oncogene, 15, 2145 ± 2150.

O'Reilly MS, Boehm T, Shing Y, Fukai N, Vasios G, Lane WS, Flynn E, Birkhead JR, Olsen BR and Folkman J. (1997). Cell, 88, 277 ± 285. O'Reilly MS, Holmgren L, Shing Y, Chen C, Rosenthal RA, Moses M, Lane WS, Cao Y, Sage EH and Folkman J. (1994). Cell, 79, 315 ± 328. Shtivelman E. (1997). Oncogene, 14, 2167 ± 2173. Shtivelman E and Namikawa R. (1995). Proc. Natl. Acad. Sci. USA, 92, 4661 ± 4665. Strieter RM, Kunkel SL, Arenberg DA, Burdick MD and Polverini PJ. (1995). Biochem. Biophys. Res. Commun., 210, 51 ± 57. Vazquez F, Hastings G, Ortega MA, Lane TF, Oikemus S, Lombardo M and Iruela-Arispe ML. (1999). J. Biol. Chem., 274, 23349 ± 23357. Whitman S, Wang X, Shalaby R and Shtivelman E. (2000). Mol. Cell. Biol., 20, 583 ± 593. Witzenbichler B, Maisonpierre PC, Jones P, Yancopoulos GD and Isner JM. (1998). J. Biol. Chem., 273, 18514 ± 18521. Xiao H, Palhan V, Yang Y and Roeder RG. (2000). EMBO J., 19, 956 ± 963. Xiao H, Tao Y, Greenblatt J and Roeder RG. (1998). Proc. Natl. Acad. Sci. USA, 95, 2146 ± 2151. Zhai Y, Ni J, Jiang GW, Lu J, Xing L, Lincoln C, Carter KC, Janat F, Kozak D, Xu S, Rojas L, Aggarwal BB, Ruben S, Li LY, Gentz R and Yu GL. (1999). FASEB J., 13, 181 ± 189.

Oncogene