Vascular permeability factor/vascular endothelial growth factor - Nature

Oncogene (1997) 14, 2025 ± 2032  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Vascular permeability factor/vascular endothelial growth factor (VPF/ VEGF) delays and induces escape from senescence in human dermal microvascular endothelial cells Yoshifumi Watanabe1,3, Sam W Lee2, Michael Detmar1, Itsuki Ajioka3 and Harold F Dvorak1,4 1

Department of Pathology and 2Department of Medicine, Gerontology Division, Beth Israel Hospital and Harvard Medical School, 330 Brookline Avenue, Boston, Massachusetts 02215, USA and 3Department of Biomolecular Engineering, Tokyo Institute of Technology, 4259 Nagatsuda, Midori-ku, Yokohama, 226, Japan

Like most other normal cells, human endothelial cells possess a limited replicative life span, and, after multiple passages in vitro, develop an arrest in cell division referred to as replicative senescence. For many cell types senescence can be delayed by oncogenes or tumor suppressor genes or prevented altogether by malignant transformation; however, once developed, senescence has been regarded as irreversible. We now report that a cytokine, vascular permeability factor/vascular endothelial growth factor (VPF/VEGF), signi®cantly delays senescence in human dermal microvascular endothelial cells (HDMEC). Typically, VPF/VEGF-treated HDMEC could be cultured for at least 15 ± 20 more population doublings (PD) than control cells. Protection from senescence was reversible in that subsequent withdrawal of VPF/VEGF returned cells to the senescent phenotype. Expression of several cell cycle-related genes (p21, p16 and p27) was signi®cantly reduced in VPF/ VEGF-treated cells but p53 expression was not signi®cantly altered. Of particular importance, VPF/VEGF was able to rescue senescent HDMEC, restoring them to proliferation, to a more normal morphology, and to reduced expression of a senescence marker, neutral bgalactosidase. Taken together, VPF/VEGF delayed the onset of senescence and also reversed senescence in microvascular endothelial cells without inducing cell transformation. Keywords: VPF/VEGF; senescence; cell cycle; endothelial cells

Introduction Normal cells have a limited capacity to divide in culture and after a ®nite number of cell divisions undergo replicative senescence (Campisi, 1996; Dimri and Campisi, 1994; Kirkland, 1992; Smith and PereiraSmith, 1996). Senescence has been regarded as an underlying cause of organismic aging, and, alternatively, as a tumor suppressive mechanism (Campisi, 1996). After repeated passages in vitro, some cells (e.g., ®broblasts) isolated from some species (e.g., rodents) Correspondence: HF Dvorak 4 Current address: Department of Biomolecular Engineering, Tokyo Institute of Technology, 4259 Nagatsuda, Midori-ku, Yokohama, 266, Japan Received 26 August 1996; revised 3 January 1997; accepted 13 January 1997

avoid senescence by undergoing transformation and immortalization. Such cellular transformation is thought to represent an important step on the pathway to neoplasia (Shay et al., 1991b). Recent studies have demonstrated that replicative senescence can be delayed in cultured cells by the expression of DNA tumor virus genes such as SV40 large T antigen (Radna et al., 1989) or HPV 16 E6 protein (Shay et al., 1993), by p53 mutation (Bond et al., 1994, 1995), by inactivation of p53 or Rb (Dimri and Campisi, 1994; Shay et al., 1991a), or, in the case of epidermal keratinocytes, by culture with epidermal growth factor (EGF) (Rheinwald and Green, 1977). However, none of these procedures has been found to reverse senescence (Smith and Pereira-Smith, 1996). The growth arrest associated with senescence is reported to be refractory to all known physiological mitogens (Dimri and Campisi, 1994) and senescent cells are thought to be terminally nondividing (Smith and Pereira-Smith, 1996). Like many other cell types, vascular endothelial cells undergo senescence when cultured in vitro (Maier et al., 1990). However, previous studies of endothelial senescence were conducted without the addition of vascular permeability factor/vascular endothelial growth factor (VPF/VEGF), a cytokine now known to be critically important in the regulation of many aspects of endothelial cell function. VPF/VEGF renders microvessels hyperpermeable to the passage of macromolecules and stimulates endothelial cells to migrate and to divide (reviewed in Dvorak et al., 1995; Senger et al., 1993; Thomas, 1996). In addition, VPF/VEGF alters endothelial cell gene expression and is critically important for both vasculogenesis and angiogenesis. All of these eects are mediated by means of high anity tyrosine kinase receptors that are expressed predominantly, though not exclusively, on vascular endothelial cells. We therefore considered the possibility that VPF/VEGF might have a role in regulating endothelial cell senescence. To test this possibility, pre-senescent microvascular endothelial cells were cultured for multiple passages with or without added VPF/VEGF. We here report that VPF/VEGF signi®cantly extended the proliferative life span of endothelial cells in vitro. We also report that VPF/VEGF reversed endothelial cell senescence, returning non-proliferating endothelial cells to a replicating phenotype. Preliminary evidence suggests that VPF/ VEGF exerted these eects, at least in part, by downregulating cell cycle-dependent kinase inhibitors.

VPF/VEGF delays/reverses senescence in HDMEC Y Watanabe et al

2026

Results Morphology and proliferation of HDMEC cultured with or without VPF/VEGF HDMEC passaged continuously without added VPF/ VEGF developed replicative senescence (de®ned as passage doubling time greater than 1 month) sometime after PD26 and before PD36 (Figure 1). Cells became

progressively larger with increasing passage number (Figure 1a ± c), a typical sign of cellular senescence (Maier et al., 1990). In addition, cell proliferation rate decreased as measured by a progressive lengthening of the passage doubling time and by a dramatic reduction in [3H]thymidine incorporation (Figure 2). Though no longer dividing, senescent cells remained viable as assessed by their capacity to exclude trypan blue, reattach to collagen-coated plates after trypsinization,

Figure 1 HDMEC were cultured without or with the addition of 5 ng/ml VPF/VEGF beginning after PD 26 (presenescence). (a) Early passage (PD10), (b) Late passage (PD30), and (c) Senescent (PD36) HDMEC cultured without VPF/VEGF. (d) HDMEC at PD 47 cultured with VPF/VEGF beginning after PD26. (e) PD45 cells pulsed with VPF/VEGF at the start of each passage after PD26, photographed 7 days following VPF/VEGF withdrawal. (f) PD56 HDMEC cultured with VPF/VEGF at each passage after PD26. Note that small proliferating cells (arrow head) and larger senescent cells co-exist in the same ®eld. Culture with VPF/VEGF delayed development of the senescent phenotype; nonetheless, even cells cultured with VPF/VEGF eventually increased in size and became senescent. All magni®cations: 6220


and maintain a level of cytoskeletal organization (Figures 1 and 3). On the other hand, when cultures were supplemented with a single pulse of VPF/VEGF at a ®nal concentration of 5 ng/ml at the time of each passage after PD26, many HDMECs continued to maintain a relatively small size even beyond PD56 (Figure 1d and f) and continued to proliferate. Thus, VPF/VEGFtreated late passage HDMEC (PD47) incorporated substantial [3H]thymidine in contrast to senescent PD36 cells cultured without VPF/VEGF which incorporated almost no radioactivity (Figure 2). VPF/VEGF's ability to protect endothelium from senescence was reversible in that when VPF/VEGF was withdrawn HDMEC developed a senescent morphology within 7 days (Figure 1e). This reversibility provides strong evidence that VPF/VEGF does not protect against senescence by causing cell transformation. In addition, we did not detect telomerase activity in VPF/VEGF-treated HDMEC (data not shown); telomerase activity has recently come to be accepted as one hallmark of cell transformation (Kim et al., 1994). We also examined the nuclear morphology of HDMEC cultured with or without VPF/VEGF. Early passage (PD 10) HDMEC have single, small, regularly oval nuclei (Figure 3A). However, HDMEC cultured to PD36 without VPF/VEGF developed a number of nuclear changes, including variable size, multinucleation and variable intensity with Hoechst dye 33342 (Figure 3B). In contrast, when HDMEC were cultured with VPF/VEGF, normal nuclear morphology was largely maintained even at PD56 (Figure 3C). In addition to its eects on endothelial nuclei, VPF/ VEGF also caused a substantial rearrangement of the cytoskeleton such that peripheral bundles of F-actin moved centrally to form prominent stress ®bers

Figure 2 Incorporation of [3H]thymidine by HDMEC cultured for dierent numbers of passage with or without VPF/VEGF as described in Materials and methods. Cell number was determined by microscopic counting so that thymidine incorporation could be normalized on the basis of cell number. PD10 cells cultured without VPF/VEGF incorporated signi®cantly more thymidine than senescent PD36 cells (P50.001). PD47 cells cultured continuously with VPF/VEGF (5 ng/ml) after PD26 incorporated thymidine much as early passage PD10 cells. However, PD45 cells cultured with VPF/VEGF continuously after PD26 and then withdrawn from VPF/VEGF for 7 days incorporated signi®cantly less thymidine than PD47 cells cultured continuously with VPF/VEGF (P50.001)

(compare right panels, Figures 3A ± C). This eect has also been observed within minutes of exposure of early passage (e.g., PD4) HDMEC to VPF/VEGF and presumably is unrelated to senescence (J Weil, unpublished data). Altered gene expression in senescent HDMEC A number of dierent cell cycle genes have been implicated in regulating cellular senescence. For example, p16/INK4/MTS1 has been shown to be involved in the cell cycle arrest of senescence (Hara et al., 1996) and p27/Kip1 has been implicated in G1 arrest (Coats et al., 1996). p53 and p21 (also designated Sdi1/WAF1/Cip1) have been extensively investigated in aging research (Bond et al., 1995; Noda et al., 1994; Shay et al., 1991a,b; Tahara et al., 1995). Also, p21 was identi®ed variously as a CDK inhibitor (Cip1) (Harper et al., 1993; Xiong et al., 1993), a tumor suppressor (WAF1) (El-Deiry et al., 1993), and as an inhibitor of DNA synthesis in senescent cells (Sdi) (Noda et al., 1994). However, there are inconsistencies in the literature. Thus, p21 has been proposed as a hallmark of cellular senescence (Noda et al., 1994) and plays a major role in G1 arrest (Deng et al., 1995). However, others have suggested that p21 is not an important determinant of senescence in that its expression was not repressed in cells which escaped from senescence by transformation (Bond et al., 1995; Tahara et al., 1995). We investigated the role of these various genes in endothelial cell senescence and their possible regulation by VPF/VEGF. As shown in Figure 4, p16 and p21 were only weakly expressed in early passage HDMEC (lane 3) but both were strongly expressed in senescent cells (lane 2). The expression of both was markedly reduced when senescence was prevented by culture with VPF/VEGF (lane 4). Withdrawal of VPF/VEGF for 7 days from HDMEC at PD45 provoked substantially increased expression of p16 (compare lanes 4 and 5), suggesting that p16 is regulated during endothelial cell senescence. However, withdrawal of VPF/VEGF had little eect on p21 expression. This ®nding is consistent with independent evidence that p21 may not be an important determinant of senescence in that its expression was not repressed in cells whose senescence was prevented by introduction of dominant mutant p53 (Bond et al., 1995; Tahara et al., 1995). p27/Kip1 was moderately expressed in senescent HDMEC but at very low or undetectable levels in both early passage cells cultured without VPF/VEGF and in late passage cells cultured with VPF/VEGF (Figure 4). Upon subsequent withdrawal of VPF/VEGF for 7 days, detectable amounts of p27/Kip1 were again expressed. Although several reports have implicated p53 in cellular senescence (Bond et al., 1994, 1995; Radna et al., 1989; Shay et al., 1993), we did not detect consistent dierences in p53 expression among early passage or senescent HDMEC nor was p53 expression altered by culture with VPF/VEGF or by subsequent VPF/VEGF withdrawal. Others have also reported that p53 expression did not change with cellular senescence (Afshari et al., 1993). However, p53 can be qualitatively and functionally modi®ed (Hall et al., 1996; Hupp et al., 1992, 1995; Wang and Prives, 1995);

2027


2028

therefore, the p53 expressed in senescent cells and VPF/VEGF-treated cells could be functionally different, a possibility that requires further investigation. Maier et al. (1994) suggested that endogenous IL-1a is critical in the progression of human umbilical vein

endothelial cells toward senescence. However, we did not detect any signi®cant dierences in IL-1a expression between young, senescent and VPF/VEGF-treated HDMEC (data not shown). These same authors reported that plasminogen activator inhibitor-1

Figure 3 Nuclear morphology of HDMEC at various passages as determined by staining with Hoechst 33342 dye alone (left panel) or with Hoechst 33342 dye plus TRITC-phalloidin (right panel). Cells were cultured for 1 day on chamber slides coated with collagen. They were then ®xed and stained as described in Materials and methods. (a) Early passage (PD10) cells. Note uniformly sized small, oval nuclei and uniform intensity of nuclear staining. (b) Senescent PD36 cells. Cells are enlarged and nuclei (arrows) show substantial variation in size and staining intensity as well as multinucleation. (c) PD47 cells cultured with VPF/VEGF after PD26. Cell size is intermediate between that of (a) and (c). Nuclei are regular, oval shaped and resemble those of (a) except that they are slightly larger. Cytoplasm contains numerous stress ®bers. All magni®cations 6170


1

2

3

4

5

p53

p21

p16

p27 Figure 4 Western blotting was performed as described in Materials and methods to evaluate the expression levels of several senescence-related genes in various passages of cultured HDMEC. Lane (1): Immortalized human keratinocytes (HaCat); lane (2): senescent PD36 cells; lane (3): Early passage PD10 cells; lane (4): PD46 cells cultured with 5 ng/ml VPF/VEGF after PD26; lane (5): PD45 cells that had been cultured continuously with VPF/VEGF after PD26, following which VPF/VEGF was withdrawn and cells were cultured for an additional 7 days

(PAI-1), induced by endogenous IL-1a (Comi et al., 1995), is a marker of senescence (Gar®nkel et al., 1994). However, VPF/VEGF is known to induce PAI-1 expression in microvascular endothelial cells (Pepper et al., 1991). Furthermore, IL-1a is thought to inhibit cell proliferation by blocking the eects of growth factors (Cozzolino et al., 1990). Taken together, it is unlikely that VPF/VEGF blocks senescence in HDMEC by regulating endogenous IL-1a levels. The discrepancy may be attributable to the diversity of response of dierent endothelial cell populations to cytokines (Barleon et al., 1996; Hitraya et al., 1995). Eects of VPF/VEGF on reversing senescence Finally, experiments were performed to determine whether VPF/VEGF could reverse replicative senescence in HDMEC. Addition of VPF/VEGF to senescent PD36 HDMEC induced renewed cell proliferation and restored a more normal morphology (Figure 5). Moreover, senescent cells subsequently cultured with VPF/VEGF reduced their expression of neutral b-galactosidase, an enzyme selectively expressed in senescent cells (Dimri et al., 1995). Whereas 79% of senescent PD36 cells stained for neutral b-galactosidase activity (Figure 5, B1), fewer than 1% of cells exhibited staining after 20 days of culture with VPF/VEGF (Figure 5, B2). In contrast, bFGF, a potent endothelial cell mitogen, did not reverse b-galactosidase staining (65% staining, Figure 5, B3) in senescent HDMEC, nor did it restore cell proliferation or normal morphology. Taken together, these data indicate that VPF/VEGF selectively reversed senescence in HDMEC. Discussion The experiments reported here have shown that VPF/ VEGF protects cultured human microvascular endothelial cells from undergong senescence without inducing malignant transformation. When HDMEC were cultured with low concentrations of VPF/VEGF

beginning after PD 26, senescence was delayed by at least an additional 15 ± 20 passage doublings. More remarkably, VPF/VEGF also restored the proliferative capacity of senescent endothelial cells while returning them to a more normal morphology and reducing their expression of neutral b-gal. To our knowledge, this is the ®rst report of a reversal of established cellular senescence. Although cellular senescence can be delayed by expression of tumor virus genes (Bond et al., 1994; Radna et al., 1989; Shay et al., 1993), by inactivation of p53 or Rb (Dimri and Campisi, 1994; Shay et al., 1991a), or in the case of epidermal keratinocytes by supplementation with EGF (Rheinwald and Green, 1977), none of these procedures was able to restore replication in cell populations with established senescence (Smith and Pereira-Smith, 1996). Even SV40 large T antigen, which does stimulate senescent cells to initiate DNA synthesis, fails to induce a complete cell cycle and such cells do not undergo mitosis (Dimri and Campisi, 1994). Do these in vitro ®ndings have any applicability to endothelial cell function in vivo? Perhaps. VPF/VEGF is strongly expressed by a number of dierent normal cells and tissues in adults in the absence of detectable angiogenesis (Berse et al., 1992). As a consequence, low levels of VPF/VEGF are found in normal plasma (Yamamoto et al., 1996), despite an extremely low level of endothelial cell turnover in adults (Engerman et al., 1967). The purpose of this cirulating VPF/VEGF has been puzzling. An obvious possibility suggested by our data is that circulating VPF/VEGF prevents vascular endothelium from undergoing replicative senescence, thereby maintaining its characteristic latent capacity for replication. This explanation could account for the fact that wounds, tumors, chronic in¯ammation, etc. elicit an angiogenic response even in aged subjects (also see note added in proof). Taken together, therefore, our data indicating that VPF/VEGF prevents (and can reverse) endothelial cell senescence in vitro suggest that VPF/VEGF might have a similar role in vivo; that is, a role in maintaining endothelial cell homeostasis including the capacity to respond to angiogenic stimulation even in aged subjects.

Materials and methods Reagents Recombinant human vascular endothelial growth factor (VEGF165) and basic ®broblast growth factor (bFGF) were purchased from R&D Systems (Minneapolis, MN). Antibodies against p16, p21, p27 and p53 were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Hoechst 33342 dye was purchased from Sigma Chemical Co, St Louis, MO. Cell culture Human dermal microvascular endothelial cells (HDMEC) were isolated from neonatal foreskins as described previously (Detmar et al., 1990). In brief, foreskins were cut into small pieces, washed in sterile phosphate-buered saline, and trypsinized overnight (0.25% trypsin, 48C) to remove the epidermis. Microvascular fragments were released by gentle scraping, ®ltered through a 100 mm sterile Nylon mesh (Reichelt Chemie-Technik, Heidelberg, Germany), washed, and puri®ed by Percoll density gradient

2029


2030

a

b

Figure 5 VPF/VEGF-mediated reversion of cellular senescence. Senescent PD36 HDMEC were cultured alone or with VPF/VEGF (50 ng/ml) or with bFGF (50 ng/ml) for 20 days and viewed with phase-contrast optics or with bright ®eld optics after staining for neutral b-galactosidase activity (blue staining). (A1, B1): Senescent PD36 HDMEC cultured without added growth factors; (A2, B2): PD36 cells cultured with VPF/VEGF; (A3, B3): PD36 cells cultured with bFGF. Note reappearance of small, cobble stone-like cells and absence of neutral b-galactosidase activity in cells cultured with VPF/VEGF (A2, B2), in striking contrast to panels (A1, B1) and (A3, B3). Magni®cations: 6220 (a), 650 (b)

centrifugation (Ruszczak et al., 1990). HDMEC were plated onto collagen type I-coated dishes at a seeding density of 26104 cells/cm2 in endothelial cell basal medium (EBM; Clonetics, San Diego, CA), supplemented with 20% fetal bovine serum (Gibco BRL, Grand Island, NY), 50 mM dibutyryl cyclic AMP, 1 mg/ml hydrocortisone acetate, 100 U/ml penicillin, 100 U/ml streptomycin and 250 mg/ ml amphotericin B (Sigma). Cultures consisted of 100% pure HDMEC, as judged by their expression of von Willebrand factor and CD31, uptake of acetylated low density lipoprotein, and expression of Weibel-Palade

bodies (Imcke et al., 1991). VPF/VEGF was undetectable (55 pM) in fetal bovine serum by immunoassay (Yeo et al., 1993). Senescent cells were generated by successive plating of 0.56105 HDMEC (starting with third passage cells) per 100 mm dish (*one-fourth con¯uence). At con¯uence (*two doublings), cells were gently trypsinized, counted, and replated and this process was repeated until cells were no longer able to undergo population doubling over the course of 1 month. Population doublings (PD) were calculated as described (Maciag et al., 1981).


To measure cell proliferation, HDMEC at dierent passages were plated at 104 cells/well in 96 well plates. After 1 day, cells were pulse labeled with [3H]thymidine (10 mCi/ml) for 4 h. Cell number was then determined by cell counting and thymidine incorporation into DNA was measured. Dierences in labeling were assessed by analysis of variance and by the Tukey ± Kramer multiple comparisons test (Zar, 1974). Cell staining and histochemistry For studies of nuclear morphology, cells were ®xed with 4% formaldehyde and stained with Hoechst dye 33342 (0.01 mM), supplemented in some cases with TRITCphalloidin for delineation of the cytoskeleton. Expression of neutral b-galactosidase, a recently described marker for senescent cells, was detected as described (Dimri et al., 1995).

with TBS, and then incubated with peroxidase-labeled second antibodies for 1 h. Bands were detected with an ECL kit (Amersham).

Note added in proof This explanation is not contradicted by our ®ndings that endothelial cells became senescent following culture in serum. At best, our cultures with 20% serum contained only one ®fth of the concentration of VPF/VEGF found in normal plasma. Also, immunoassay failed to detect VPF/ VEGF in the fetal calf serum used in our studies. This could re¯ect a lack of cross-reactivity of our anti-human VPF/VEGF antibodies with bovine VPF/VEGF or, alternatively, a loss of VPF/VEGF during the course of serum preparation. Finally, it is uncertain whether bovine VPF/ VEGF, even if present, would react with full competence on human endothelial cells.

Western blotting Cells were harvested and lysed in SDS ± PAGE sample buer and boiled for 5 min. Lysates prepared from equivalent numbers of cells (56104 cells/lane) were subjected to SDS ± PAGE in 4 ± 15% gradient gels. Proteins were then electroblotted onto PVDF membranes (Amersham, Illinois) in cold transfer buer for 3 h at 60 V. Filters were incubated with ®rst antibodies for 1 h at room temperature after blocking with 3% skim milk, washed

Acknowledgements We thank Drs Corinne L Reimer and Jan Vijg for reading the manuscript. An abstract of this paper was presented at the annual meeting of The American Society for Cell Biology, December, 1996. This work was supportd by US Public Health Service NIH grants CA-50453 and CA58845.

References Afshari CA, Vojta PJ, Annab LA, Futreal PA, Willard TB and Barrett JC. (1993). Exp. Cell. Res., 209, 231 ± 237. Barleon B, Sozzani S, Zhou D, Weich HA, Mantovani A and Marme D. (1996). Blood, 87, 3336 ± 3343. Berse B, Brown LF, Van de Water L, Dvorak HF and Senger DR. (1992). Mol. Biol. Cell., 3, 211 ± 220. Bond JA, Blaydes JP, Rowson J, Haughton MF, Smith JR and Wynford-Thomas D. (1995). Cancer Res., 55, 2404 ± 2409. Bond JA, Wyllie FS and Wynford-Thomas D. (1994). Oncogene, 9, 1885 ± 1889. Campisi J. (1996). Cell, 84, 497 ± 500. Coats S, Flanagan WM, Nourse J and Roberts JM. (1996). Science, 272, 877 ± 880. Comi P, Chiaramonte R and Maier JAM. (1995). Exp. Cell. Res., 219, 304 ± 308. Cozzolino F, Torcia M, Aldinucci D, Ziche M, Almerigogna F, Bani D and Stern DM. (1990). Proc. Natl. Acad. Sci. USA, 87, 6487 ± 6491. Deng C, Zhang P, Harper JW, Elledge SJ and Leder P. (1995). Cell, 82, 675 ± 684. Detmar M, Imcke E, Ruszczak Z and Orfanos CE. (1990). J. Invest. Dermatol., 95, 219S ± 222S. Dimri GP and Campisi J. (1994). Cold Spring Harb. Symp. Quant. Biol., 59, 67 ± 73. Dimri GP, Lee X, Basile G, Acosta M, Scott G, Roskelley C, Medrano EE, Linskens M, Rubeli I, Pereira-Smith O and Peacocke M. (1995). Proc. Natl. Acad. Sci. USA, 92, 9363 ± 9367. Dvorak HF, Brown LF, Detmar M and Dvorak AM. (1995). Am. J. Pathol., 146, 1029 ± 1039. El-Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons R, Trent JM, Lin D, Mercer WE, Kinzier KW and Vogelstein B. (1993). Cell, 75, 817 ± 825. Engerman RL, Pfaenbach D and Davis MD. (1967). Lab. Invest., 17, 738 ± 743. Gar®nkel S, Brown S, Wessendorf JHM and Maciag T. (1994). Proc. Natl. Acad. Sci. USA, 91, 1559 ± 1563.

Hall SR, Campbell LE and Meek DW. (1996). Nucl. Acid Res., 24, 1119 ± 1126. Hara E, Smith R, Parry D, Tahara H, Stone S and Peters G. (1996). Mol. Cell. Biol., 16, 859 ± 867. Harper JS, Adami GR, Wei N, Keyomarsi K and Elledge SJ. (1993). Cell, 75, 805 ± 816. Hitraya EG, Tan EML, Rudnicka L and Jimenez SA. (1995). Lab. Invest., 73, 393 ± 402. Hupp TR, Meek DW, Midgiey CA and Lane DP. (1992). Cell, 71, 875 ± 886. Hupp TR, Sparks A and Lane DP. (1995). Cell, 83, 237 ± 245. Imcke E, Ruszczak Z, Mayer-da Silva A, Detmar M and Orfanos CE. (1991). Arch Dermatol. Res., 283, 149 ± 157. Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, Ho PLC, Coviello GM, Wright WE, Weinrich SL and Shay JW. (1994). Science, 266, 2011 ± 2015. Kirkland JL. (1992). Clin. Biochem., 25, 61 ± 75. Maciag T, Hoover GA, Stemerman MB and Weinstein R. (1981). J. Cell. Biol., 91, 420 ± 426. Maier JAM, Statuto M and Ragnotti G. (1994). Mol. Cell. Biol., 14, 1845 ± 1851. Maier JAM, Voulalas P, Roeder D and Maciag T. (1990). Science, 249, 1570 ± 1574. Noda A, Ning Y, Venable SF, Pereira-Smith OM and Smith JR. (1994). Exp. Cell. Res., 211, 90 ± 98. Pepper MS, Ferrara N, Orci L and Montesano R. (1991). Biochem. Biophys. Res. Commun., 181, 902 ± 906. Radna RL, Caton Y, Jha KK, Kaplan P, Li G, Traganos F and Ozer HL. (1989). Mol. Cell. Biol., 9, 3093 ± 3096. Rheinwald JG and Green H. (1977). Nature, 265, 421 ± 424. Ruszczak Z, Detmar M, Imcke E and Orfanos CE. (1990). J. Invest. Dermatol., 95, 693 ± 699. Senger D, Van De Water L, Brown L, Nagy J, Yeo K-T, Yeo T-K, Berse B, Jackman R, Dvorak A and Dvorak H. (1993). Cancer Metasta. Rev., 12, 303 ± 324. Shay JW, Pereira-Smith OM and Wright WE. (1991a). Exp. Cell. Res., 196, 33 ± 39.

2031


2032

Shay JW, Wright WE, Brasiskyte D and Haegen BAVD. (1993). Oncogene, 8, 1407 ± 1413. Shay JW, Wright WE and Werbin H. (1991b). Biochim. Biophys. Acta, 1072, 1 ± 7. Smith JR and Pereira-Smith OM. (1996). Science, 273, 63 ± 67. Tahara H, Sato E, Noda A and Ide T. (1995). Oncogene, 10, 835 ± 840. Thomas KA. (1996). J. Biol. Chem., 271, 603 ± 606. Wang Y and Prives C. (1995). Nature, 376, 88 ± 91. Xiong Y, Hannon GJ, Zhang H, Casso D, Kobayashi R and Beach D. (1993). Nature, 366, 701 ± 704.

Yamamoto Y, Toi M, Kondo S, Matsumoto T, Suzuki H, Kitamura M, Tsuruta K, Taniguchi T, Okamoto A, Mori T, Yoshida M, Ikeda T and Tominaga T. (1996). Clin. Cancer Res., 2, 821 ± 826. Yeo K-T, Wang HH, Nagy JA, Sioussat TM, Ledbetter SR, Hoogewerf AJ, Zhou Y, Masse EM, Senger DR, Dvorak HF and Yeo T-K. (1993). Cancer Res., 53, 2912 ± 2918. Zar JH. (1994). Biostatistical Analysis. Prentice-Hall, Inc, Englewood Clis, NJ.