Multiple Stages of Malignant Transformation of Human

Multiple stages of malignant transformation of human endothelial cells modelled by co-expression of telomerase reverse transcriptase, SV40 T antigen and oncogenic N-ras

Karen L MacKenzie*,1,3, Sonia Franco1, Afzal J Naiyer2, Chad May1, Michel Sadelain1, Shahin Ra®i2 and Malcolm AS Moore1

1James Ewing Laboratory of Developmental Haematopoiesis and Department of Human Genetics, Sloan-Kettering Cancer Institute, New York, NY, USA; 2Cornell University Medical College, New York, NY, USA

We have modelled multiple stages of malignant Introduction transformation of human endothelial cells (ECs) by overexpressing the catalytic subunit of human telomerase Malignant transformation is a multi-step process, (hTERT), together with SV40 T antigen (SV40T) and during which accumulating genetic and epigenetic oncogenic N-ras. Transfection with hTERT alone, led to alterations drive normal cells to a cancerous phenotype the immortalization of two out of three cultures of bone (Fearon and Vogelstein, 1990; Land et al., 1983). marrow-derived ECs (BMECs). One hTERT transduced Phenotypic characteristics of malignant cells include an BMEC culture underwent a long proliferative lag before unlimited proliferative potential, reduced growth factor resuming proliferation. BMECs transfected with hTERT requirements, changes in antigen expression, loss of alone were functionally and phenotypically normal. specialized cell functions, changes in cell motility and BMECs transfected with SV40T (BMSVTs) had an ability to metastasise. Molecular changes that con- extended lifespan, but eventually succumbed to crisis. tribute to the transformation process include activation BMSVTs exhibited a partially transformed phenotype, of the enzyme telomerase, as well as mutation, demonstrating growth factor independence, altered ampli®cation and rearrangement of proto-oncogenes antigen expression and forming tiny, infrequent colonies and deletion or silencing of tumour suppressor genes in vitro. Transduction of BMSVTs with hTERT resulted (Kim et al., 1994; Vogelstein et al., 1988). Little is in immortalization of 4 out of 4 cultures. BMSVTs known of the molecular mechanisms that underlie immortalized with hTERT formed large colonies in vitro aberrant growth and malignant transformation of and small transient tumours in vivo. BMECs co- endothelial cells (ECs). EC-derived neoplasms include expressing SV40T, hTERT and N-ras exhibited an haemangioma, haemangioblastoma and angiosarcoma overtly transformed phenotype; forming very large (Couch et al., 2000; Drolet et al., 1999; Mark et al., colonies with an altered morphology and generating 1996). Excessive proliferation of ECs also contributes rapidly growing tumours in vivo. These investigations to malignant progression of tumours of non-EC origin demonstrate transformation of human ECs to an overtly (Bergers et al., 1999). The development of experimental malignant phenotype. This model will be useful for models of EC carcinogenesis will enable more rigorous understanding mechanisms underlying vascular and investigation of these processes. angiogenic neoplasias, as well as for testing drugs The enzyme telomerase is expressed in most cancer designed to curtail aberrant EC growth. cells and immortalized tumour cell lines (Kim et al., Oncogene (2002) 21, 4200 ± 4211. doi:10.1038/sj.onc. 1994). In contrast, normal human somatic cells do not 1205425 express telomerase and have a tightly regulated replicative lifespan (Hay¯ick and Moorhead, 1961; Keywords: telomerase; oncogenes; transformation; im- Kim et al., 1994). These observations led to the mortalization; endothelial cells proposal that telomerase activity is responsible for the immortal phenotype of cancer cells. Telomerase is a ribonuclear protein complex that has a reverse transcriptase (hTERT) as a catalytic domain and an RNA component that functions as a template for transcription. Telomerase synthesises TTAGGG DNA repeats at chromosomal termini, which are referred to as telomeres (Moyzis et al., 1988). In the absence of *Correspondence: KL MacKenzie; telomerase activity, telomeres in normal somatic cells E-mail: [email protected] shorten with each cell division (Harley et al., 1990). 3 Current address: Children's Cancer Institute Australia for Medical When telomeres become critically short, normal cells Research, High St., P.O. Box 81 Randwick, NSW, 2031 Australia Received 11 October 2001; revised 30 January 2002; accepted 20 undergo an irreversible growth arrest, referred to as February 2002 senescence. During senescence the cyclin dependent- Malignant transformation of human endothelial cells KL MacKenzie et al 4201 kinase inhibitors p21CIP1/WAF1 and p16INK4a, which assayed for EC characteristics and properties of function in pathways of the tumour suppressor genes malignant transformation. Our investigations provide p53 and RB respectively, are upregulated (Smith and a detailed model of multi-step transformation of Pereira-Smith, 1996). human ECs, demonstrating gradual loss of EC Recent studies, including our own, have demon- function and co-incidental acquisition of malignant strated that over expression of hTERT reconstitutes characteristics. telomerase activity in vivo, elongates telomeres and thereby enables primary human ®broblasts and retinal epithelial cells to proliferate beyond senescence (Bod- Results nar et al., 1998; MacKenzie et al., 2000). Transforma- tion with viral oncogenes, such as SV40 T antigen Gene transfer and expression (SV40T) also enables primary human cells to proliferate beyond senescence (Girardi et al., 1965). Expression To investigate the eects of hTERT expression on EC of SV40 Large T antigen inactivates p53 and RB lifespan, bone marrow derived ECs (BMECs), periph- pathways and thereby enables telomerase negative cells eral blood derived ECs (PBECs) and BMSVTs to bypass senescence and proliferate for an additional (BMECs that were transformed with SV40T) were 20 ± 40 population doublings (PDLs) (Shay et al., transduced with the retroviral vector MTIG, which 1991). However, cells that bypass senescence in this encodes hTERT and GFP. We have previously shown way are still subject to telomeric shortening, continue that transduction with MTIG reconstitutes telomerase to age, accumulate chromosomal abnormalities and activity, elongates telomeres and extends the lifespan of eventually succumb at a second mortality check point, primary human ®broblasts (MacKenzie et al., 2000). referred to as crisis (Counter et al., 1992; Girardi et al., The MGFP retroviral vector, which encodes the 1965). As a very rare event, a clone may escape crisis enhanced green ¯uorescence protein (GFP) only, was and proliferate inde®nitely. The frequency that SV40T employed as a control. PBECs, BMECs and BMSVTs transfected human ®broblasts escape crisis was deter- transduced with MTIG were referred to as PBhTERTs, mined to be 361077, whereas the rate that human BMhTERTs and BMSVThTERTs respectively. Simi- epithelial cells escaped crisis was about 100-fold higher larly, cells transduced with MGFP were referred to as (Shay et al., 1993). Recent studies have demonstrated PBGFPs, BMGFPs and BMSVTGFPs respectively. that induction of telomerase is critical at crisis; cells Fluorescence activated cell scanner (FACS) analysis for that spontaneously escaped crisis were shown to GFP expression demonstrated that the MTIG vector express telomerase (Klingelhutz et al., 1994) and transduced up to 40% of target BMECs, while the ectopic expression of hTERT overcame crisis and MGFP transduction eciency ranged from 35 ± 79%. immortalized virally transformed ®broblasts and Telomerase activity was detected in all MTIG- epithelial cells (Counter et al., 1998; Halvorsen et al., transduced ECs at all time points tested in several 1999; Zhu et al., 1999). Previous studies indicate that independent experiments (Table 1). High levels of immortalization is necessary, but not sucient for telomerase activity were sustained in cultures that were malignant transformation. This notion was exempli®ed assayed after more than 600 days of culture. It should by a recent study, which demonstrated that human be noted that the values shown in Table 1 for ®broblasts and epithelial cells immortalized by SV40T telomerase activity at early time points following plus hTERT were non-tumorigenic until co-transfected retroviral transduction underestimate telomerase activ- with oncogenic H-ras (Hahn et al., 1999). ity per transduced cell, since telomerase activity shown Ras proteins are GTPases that operate as GDP/GTP in Table 1 is for the entire mass culture (which includes molecular switches in a multitude of signal transduc- non-transduced cells). The level of telomerase activity tion pathways (Pruitt and Der, 2001). Ras signal detected at initial time points re¯ects the MTIG transduction pathways have diverse functional con- transduction eciency, which varied between experi- sequences, including eects on cell cycle progression ments. The apparent increase in telomerase activity at and dierentiation. In cancer cells, speci®c Ras point later time points is most likely due to selection for mutations constitutively activate signalling pathways. MTIG-transduced BMECs and BMSVTs at senescence The oncogenic eects of mutationally activated Ras are and crisis respectively. Negligible telomerase activity diverse and appear to be cell type speci®c. Oncogenic was detected in mock-transduced, MGFP-transduced changes that were previously attributed to Ras include BMECs and BMSVTs at all time points tested (Figure morphologic alterations, abrogation of growth factor 1a, Table 1). requirements, anchorage independent growth, induc- BMECs that were transfected with hTERT, SV40T tion of vascular endothelial growth factor (VEGF) and and N-ras were generated by retroviral transduction of increased cell motility (Andrejauskas and Moroni, the cell line BMSVThTERT-4 with the vector LN-ras2 1989; Fox et al., 1994; Rak et al., 1995; Seeburg et (MacKenzie et al., 1999). The resultant cell line was al., 1984). termed BMSVThTERT-4Nras. Since LN-ras2 was In the present study, we have investigated the packaged into an ecotropic Moloney murine leukaemia potential for hTERT, SV40T and oncogenic Ras to virus (M-MLV) envelope that does not infect human transform human ECs. Primary ECs were sequentially cells, the target cells were ®rst transduced with an transduced with these three genetic elements and adenoviral vector (AdEcoR) that expresses the receptor

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4202 Table 1 Telomerase activity and replicative lifespan Exogenous genes Telomerasea Lifespan Cells hTERT SV40T % Days Fold increaseb Statusc

PBMOCK-1 770 7 1.0 Normal 092 PBMOCK-2 771 3 1.0 Normal PBGFP-1 770 7 1.1 Normal 086 PBGFP-2 770 3 0.8 Normal PBhTERT-1 + 7 21 7 1.1 Normal 66 78 PBhTERT-2 + 7 17 7 0.9 Normal 13 86 PBhTERT-3 + 7 15 3 0.8 Normal BMMOCK-1 774 4 1.0 Normal BMMOCK-2 774 4 1.0 Normal BMGFP 772 4 1.0 Normal BMhTERT-1 + 7 100 4 44.3 Immortal 161 32 143 500 BMhTERT-2 + 7 31 4 44.2 Immortal 12 32 139 636 BMhTERT-3 + 7 89 4 1.3 Extended BMSVTMOCK-1 7 + 2 4 1.0 Extended BMSVTMOCK-2 7 + 1 17 1.0 Extended BMSVTMOCK-3 7 + 1 92 1.0 Extended BMSVTGFP-1 7 + 1 17 1.0 Extended BMSVTGFP-2 7 + 3 92 1.0 Extended BMSVThTERT-1 + + 37 4 426.2 Immortal 224 569 BMSVThTERT-2 + + 13 4 421.9 Immortal 157 32 300 481 BMSVThTERT-3 + + 6 17 49.1 Immortal 152 237 BMSVThTERT-4 + + 274 206 48.0 Immortal

aTelomerase activity is indicated as a percentage of a control tumour cell line. Days indicate the time that cultures were assayed for telomerase activity. Days are counted from the day of retroviral infection. bFold increase in replicative lifespan was calculated as the ratio of the maximum number of PDLs achieved (or PDLs at the time of writing): number of PDLs at which control cells senesced (BMEC-derived cells) or went into crisis (BMSVT-derived cells). cReplicative lifespan was considered extended when cells continued to proliferate beyond a 1.2-fold increase in the lifespan of control cells. The lifespan of all BMSVT cells was extended by virtue of transformation with SV40T

for ecotropic M-MLV (Scott-Taylor et al., 1998). As a arrest (data not shown). In contrast to BMhTERT-1, control, ECs were transduced with AdEcoR and then the BMhTERT-2 culture bypassed senescence and then ecotropic MGFP. FACS analysis for GFP expression underwent a long lag phase before resuming prolifera- demonstrated that 49% of the cells were transduced tion. The BMhTERT-3 culture exhibited a very modest with the MGFP vector. Expression of N-ras in increase in lifespan (1.3-fold) before proliferation was BMSVThTERT-4Nras cells was con®rmed by RT ± permanently arrested. The morphology of growth PCR (Figure 1b). arrested BMhTERT-3 cells was characteristic of senescent cells. To date, BMhTERT-1 and BMhTERT-2 cultures continue to proliferate after Effect of hTERT expression on EC lifespan more than 600 days in culture and a greater than Although retroviral transduction of primary PBECs fourfold increase in lifespan. We therefore conclude with MTIG induced telomerase activity, there was no that the BMhTERT-1 and BMhTERT-2 cells are signi®cant eect on the replicative lifespan of three immortal. However, the variety of growth patterns PBEC cultures (PBhTERT-1,2,3, Table 1). In contrast, observed for the three BMhTERT cultures is consistent MTIG transduction extended the lifespan of three with the requirement for an additional molecular event cultures of primary BMECs (Figure 2a). However, for immortalization. dierent growth kinetics resulted from three indepen- The relative resistance of primary PBECs and dent BMEC transductions. The BMhTERT-1 culture BMECs to hTERT-mediated immortalization con- readily bypassed senescence and continued to prolifer- trasts with a previous investigation that reported ate, albeit at a slower rate. The reduced growth rate of one-step hTERT-driven immortalization of ECs de- post-senescent BMhTERT-1 cells appeared to be due rived from alternative tissue sources (Yang et al., subsets of cells undergoing a senescent-like growth 1999). A recent investigation suggested that hTERT-

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4203

Figure 1 Gene transfer and expression in transduced ECs. (a) Telomeric repeat ampli®cation protocol (TRAP) assay for telomerase activity. The arrow indicates an internal control used for quanti®cation of telomerase activity. The days indicate the number of days since transduction with hTERT. (b) RT ± PCR for expression of LNras2 vector and b-actin. Lane 1; viral producer cells (cN-ras2), lane 2 viral producer cells without reverse transcriptase, lanes 3 ± 5; cell lines derived from subcutaneous tumours (A and B), and a lung tumour that grew in mice injected with the cell line BMSVThTERT-4Nras, lane 6; BMSVThTERT- 4Nras cells, lane 7; BMSVThTERT-4 cells, lane 8; H2O mediated immortalization is hampered by inadequate culture conditions (Ramirez et al., 2001). We therefore compared our culture media (ECGM, described in Materials and Methods) with the media employed by Yang et al. (1999) (EBM2, described in Materials and Figure 2 Eects of hTERT expression on BMEC lifespan. (a) Methods) to immortalize ECs. Non-transduced Proliferation of primary BMECs that were either mock-infected BMECs, BMhTERT-1 and BMhTERT-2 cells that (BMMOCK-1 and BMOCK-2), or transduced with either MGFP were cryopreserved at early passage were thawed and (BMGFP) or MTIG (BMhTERT-1-3). (b) Comparison of split into either EBM2 or ECGM. As shown in Figure proliferation of control (non-transduced) BMECs and hTERT- 2b, the growth of control BMECs and BMhTERT transduced BMECs when grown in ECGM versus EBM2. (c) Proliferation of BMSVT cells that were either mock-infected cells was unaltered by freeze/thawing and propagation (BMSVTMOCK-1) or transduced with either MGFP in the alternative media. Control BMECs senesced (BMSVTGFP-2) or MTIG (three dierent BMSVThTERT cell after the same number of population doublings in the lines are shown). BMSVThTERT-4Nras are BMSVTs transduced two dierent media. Furthermore, the growth curves with both MTIG and LN-ras2. All cultures are post-senescence. for BMhTERT-1 and BMhTERT-2 were identical Growth arrest of control cells at 25 PDLs represents crisis under both culture conditions. These results strongly suggest that the apparent resistance to immortalization spontaneously immortalized cell lines have arisen from displayed by BMECs and PBECs compared to ECs BMSVT cultures. However, transduction with MTIG from other sources was not due to adverse culture very eectively overcame crisis and immortalized four conditions. out of four BMSVThTERT cultures (Figure 2c and As a consequence of transfection with SV40T, the Table 1). Indeed, BMSVThTERT cultures overcame starting population of BMSVTs bypassed senescence crisis even when the MTIG transduction was very (Candal et al., 1996). BMSVT cells proliferated for inecient (estimated at 1% by FACS analysis in the 25 ± 30 PDLs before succumbing to crisis. Thus far, no case of BMSVThTERT-3). The most advanced

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4204 BMSVThTERT culture has undergone more than 500 (VE-CADH). Uptake of Diacetylated low density PDLs, which equates to more than a 26-fold increase lipoprotein (Dil-Ac-LDL) was also assessed by FACS in lifespan (Table 1). analysis (Table 2). Expression of von Willibrand factor VIII (VWF) was assessed by immunohisto- chemical staining of cells grown on chamber slides. Telomere dynamics in BMEC culture The results summarized in Table 2 describe a normal Telomere length was measured by Southern blot phenotype for BMhTERT cells that were assayed analysis of telameric restriction fragment (TRF) lengths between 54 and 121 PDLs. The only dierence (Figure 3). As expected, the mean and peak TRF between control BMECs and BMhTERT cells was length of non-transduced BMECs shortened as the cells reduced expression of CD34 on some BMhTERT were propagated and approached senescence. Mean cells. However, CD34 was also down regulated on TRF measurements of non-transduced BMECs shor- non-transduced senescent BMECs (PDL 29). BMECs tened from about 12 Kbp, at the earliest time point and BMhTERT cells were also indistinguishable with measured to around 7.0 Kbp at senescence. Despite regard to ability to support haematopoiesis and form high levels of telomerase, telomeres also shortened in angiogenic webs in vitro (KL MacKenzie, AJ Naiyer, BMhTERT-1 cells, albeit at a slower rate. The S Ra®i, MAS Moore, unpublished data). In contrast telomeres of BMhTERT-1 cells eventually stabilized to control BMECs and BMhTERT cells, the pheno- around 7.0 Kbp (Figure 3b). Telomeres of BMSVT type of BMSVTs was considerably altered. BMSVTs cells shortened to 6.0 Kbp at crisis. Induction of expressed lower levels of KDR and P-CAM than telomerase in BMSVThTERT cells resulted in main- BMhTERT cells, even though BMhTERTs were tenance of telomeres between 4 and 6 Kbp (Figure assayed at higher PDLs. VWF was expressed by early 3a,b). The telomeres of BMSVThTERT cells trans- passage BMSVTs, but not BMSVTs approaching duced with N-ras ¯uctuated within 4 ± 7 Kbp (data not crisis. Uptake of Dil-Ac-LDL was also reduced on shown). BMSVT cells. Analysis of BMSVThTERT cells, which were assayed after more than 200 PDLs, revealed further deviation from normal BMECs, with complete BMEC phenotype loss of expression of VE-CADH, P-CAM and VWF, FACS analysis was performed to quantify expression as well as inability of some BMSVThTERT cell lines of the cell surface markers KDR (VEGF receptor-2), to take up Dil-Ac-LDL. The phenotype of the triple CD34, I-CAM, P-CAM (CD31) and VE-Cadherin transfected cell line, BMSVThTERT-4Nras, was similar to BMSVThTERT-4 cells. BMhTERT, BMSVT and BMSVThTERT cells retained a normal EC morphology, exhibited typical polyhedral morphology and grew in monolayers in cobblestone patterns. In contrast, the BMSVThTERT- 4Nras cell line demonstrated a less organized growth pattern, with cells overlaying one another. In addition, BMSVThTERT-4Nras cells had an irregular, more spindle shaped morphology and were highly refractile (data not shown).

Malignant transformation of BMECs BMSVT, BMSVThTERT cells and BMSVThTERT- 4Nras cells grew independently of exogenous VEGF and grew to a higher cell density than BMECs and BMhTERT cells (data not shown). In vitro clonogenic assays were performed to assess anchorage independent growth of all BMECs (Table 3, Figure 4). The HT1080 ®brosarcoma cell line was employed as a positive control for colony growth. No colonies or clusters were detected in agarose cultures of early passage BMECs (13 ± 17 PDLs). BMhTERTs, which were assayed at a number of time points after 50 PDLs also failed to form colonies. Colony formation by BMSVT cells was Figure 3 Telomere regulation in hTERT-transduced BMECs. (a) assessed in several independent tests, from 0 ± 25 PDLs TRF analysis of BMEC telomere length. White bars indicate the prior to crisis. BMSVTs formed many tiny clusters of mean TRF length. Molecular weight markers are indicated in Kbp less than 50 cells. Larger BMSVT clusters of a few at the left of the gels. The number of PDLs at the time of sampling are indicated at the top of the gel. (b) Graphic representation of hundred cells were detected at a low frequency of 0.4% changes in mean telomere length in transduced BMECs. Values and scored as colonies (Table 3 and Figure 4a). were calculated as averages of TRF results from several gels BMSVThTERT cells were assayed for anchorage

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4205 independent growth at various time points from 70 size of 0.7 cm in diameter 14 days after injection, then PDLs onward. All four BMSVThTERT cell lines regressing by day 21. Mice injected with formed large colonies at frequencies that ranged from BMSVThTERT cells remained healthy for the duration 4 to 17% of the total cells plated. Variations in the of the experiment. In contrast, all ®ve mice (3 out of 3 cloning eciencies among the four BMSVThTERT cell Nude and 2 out of 2 NOD/SCID) injected with lines were not statistically dierent (Student's t-test). BMSVThTERT-4Nras cells developed rapidly growing BMSVThTERT colonies exhibited a very regular, and sustainable tumours. The BMSVThTERT-4Nras spherical or elliptical morphology. Cells were tightly tumours reached 1.5 ± 2.0 cm in diameter within 3 packed within a basement membrane that surrounded weeks of injection. These mice were sacri®ced due to the colony (Figure 4b). BMSVThTERT-4Nras cells tumour burden and deteriorating health. A metastatic formed colonies with a similar eciency as the growth was detected at a lymph node of one Nude BMSVThTERT-4 cell line. However, compared with mouse injected with BMSVThTERT-4Nras cells. BMSVThTERT-4 colonies, there was a higher in- Examination of histologic sections of tumours and cidence of very large colonies, that measured greater metastases revealed large polyhedral to spindle shaped than 0.5 mm in diameter, in the BMSVThTERT-4Nras cells that were organized into lumen like structures and cultures. Furthermore, subsets of the colonies gener- capillary network patterns in some areas (Figure 4d). ated by the N-ras transduced cells exhibited an Many mitotic ®gures were apparent among the tumour irregular morphology as shown in Figure 4c. cells. Tumours also grew in mice injected with HT1080 To assess tumorigenicity, immunocompromised mice cells. However, the histology of HT1080 tumours was received subcutaneous (SC) injections of 5 ± clearly distinct from BMSVThTERT-4Nras tumours 106106 cells. SC tumour formation was assessed in (data not shown). To further assess the malignant both Nude mice and sub-lethally irradiated NOD/ potential of the transformed ECs, BMSVThTERT- SCID mice. HT1080 cells were employed as controls. 4Nras cells were injected via the tail vein of sublethally- Over 12 ± 16 weeks of observation, no tumours irradiated NOD/SCID mice. All three mice injected by developed in mice that were injected with control intravenous (IV) route became moribund and were BMEC, BMhTERT or BMSVT cells. Eighty per cent sacri®ced within 3 weeks of injection. Upon autopsy of mice (5 out of 7 Nude and 3 out of 3 NOD/SCID) and histologic examination, it was evident that the EC that were injected with BMSVThTERT-1 cells, and tumour cells had lodged and grown within the lungs. 20% of mice (0 out of 3 Nude and 1 out of 2 NOD/ There was no histologic evidence of EC tumour growth SCID) injected with BMSVThTERT-4 cells developed in the spleen, liver, bone marrow and peripheral blood. small, SC growths at the site of injection. However, Cell lines were readily established by culturing cell these tumours were transient, growing to a maximum suspensions from SC tumours and the lungs of mice

Table 2 BMEC phenotypes Cells KDR CD34 I-CAM P-CAM VE-CADH VWF Dil-Ac-LDL

BMEC (early passage) + + + ++ + + ++ BMEC (late passage) 77++ ++ + + ++ BMhTERT + +/7 ++++ + ++ BMSVT +/7 +/7 ++ ++/7 + BMSVThTERT +/7 +/7 + 777+/7 BMSVThTERT-4Nras 7 NE NE 7 NE 7 +

+or7 indicates antigen expression as determined by FACS analysis of mean ¯ourercence. ++, strong postive, 20 ± 200-fold increase in mean ¯uorescence; +, weak to medium positive, 2 ± 20-fold increase in mean ¯uorescence; 7, negative, less than twofold increase in mean ¯uoresence; +/7, some cell lines weak positive, others negative, NE, not evaluated

Table 3 Colonies and tumours Tumours Transduced genes Colonies Subcutaneous (SC)b Cells SV40T hTERT N-ras Per 10 000 cells n Regressed Established IV

BMEC 777 0+0 2 NE NE NE BMhTERT-1 7 + 7 0+0 3 0/3 0/6 NE BMSVTMOCK-2 + 77 41+37 7 0/8 0/12 NE BMSVThTERT-1 + + 7 1691+756 9 8/10 0/14 NE BMSVThTERT-2 + + 7 391+222 3 NE NE NE BMSVThTERT-3 + + 7 483+404 2 NE NE NE BMSVThTERT-4 + + 7 1374+927 5 1/5 0/5 NE BMSVThTERT-4Nras + + + 1619+913a 6 NA 5/5 3/3 HT1080 NA NA NA 2402+1017 2 NA 7/7 3/3 aColony morphology was altered compared with BMSVThTERT colonies. bIncludes analysis of SC tumours in both Nude and NOD/SCID mice. NA, not applicable. NE, not evaluated

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4206 receiving IV injections of BMSVThTERT-4Nras cells. Discussion Cell lines that were established from BMSVThTERT- 4Nras tumours had identical morphology, phenotype Replicative lifespan of BMECs expressing hTERT and and growth characteristics as the injected parental cells. SV40T Expression of the LN-ras2 vector was detected in cell lines grown from tumour tissue (Figure 1b). Two prior studies of hTERT transfected human ECs have reported con¯icting results. While one study indicated that expression of hTERT is sucient for immortalization (Yang et al., 1999), the other study demonstrated that hTERT alone does not immortalize human ECs (O'Hare et al., 2001). Yang et al. (1999) demonstrated that the lifespan of ECs transfected with hTERT was increased by 2.5- to ®vefold, which was interpreted as immortalization. In the more recent study by O'Hare et al. (2001), ECs transduced with hTERT exhibited a very modest lifespan extension (1.4-fold) before undergoing irreversible growth arrest. In the present investigation, a range of proliferative responses, which encompassed both of these previous reports, were observed when BMECs and PBECs were transduced with hTERT. Transduction with hTERT increased the lifespan of 3 out of 3 cultures of BMECs, but had no signi®cant eect on the lifespan of PBEC cultures. Out of the three BMhTERT cultures with an extended lifespan, only two appeared to be immortalized. However, the variability in the proliferative response of the three hTERT-transduced BMEC cultures and the long lag period exhibited by BMhTERT-2, suggests that additional spontaneous events were necessary for immortalization. It may be signi®cant that the BMhTERT-1 culture, which bypassed senescence without any apparent proliferative lag, was transduced with the highest eciency. The larger number of transduced cells within the BMhTERT-1 mass culture would have provided the greatest opportunity for a clone or subset of cells to sustain immortalizing mutations. Under standard culture conditions, induction of telomerase activity alone was not sucient for immortalization of mammary epithelial cells or keratinocytes (Dickinson et al., 2000; Kiyono et al., 1998). A recent study has suggested that this apparent resistance to immortalization was due to inadequate culture conditions (Ramirez et al., 2001). However in the present study, BMECs were equally resistant to immortalization with hTERT whether cultured in ECGM or grown under the conditions shown by Yang et al. (1999) to be adequate for immortalization. Furthermore, our results demonstrate three dierent growth patterns for BMhTERT cells cultured under identical conditions. These results strongly suggest that culture conditions were not responsible for variations in the proliferative response of ECs transduced with hTERT. However, we cannot rule out the possibility that minor variations in cell culture technique contributed to the various outcomes observed in dierent Figure 4 Malignant transformation of BMECs. Photomicro- laboratories. Another plausible explanation for the graphs of BMEC colonies grown in soft agar (magni®cations apparent discrepancies is that ECs derived from are 506); (a) BMSVT, (b) BMSVThTERT-4, (c) various tissue origins dier in susceptibility to im- BMSVThTERT-4Nras. (d) Histologic section of a SC tumour in mortalization with hTERT. While BMECs and PBECs a nude mouse injected with BMSVThTERT-4Nras cells (magni- ®cation 9006). Shows mitotic tumour cells (white arrow) and were utilized in the present investigations, O'Hare et al. lumen or capillary-like structures (black arrow) (2001) employed mammary microvascular ECs and

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4207 Yang et al. (1999) transduced human ECs derived from telomerase may conceal eroded chromosomal ends umbilical vein, saphenous vein, aorta and neonatal from DNA damage machinery and thereby protect foreskin. The possibility that EC strains dier in the cells from growth arrest or apoptosis. In our susceptibility to immortalization is consistent with previous study, MTIG transduction of primary ®bro- previous studies which demonstrated that under blasts led to rapid elongation of telomeres, from 7 up standard culture conditions, hTERT expression is to 30 Kbp (MacKenzie et al., 2000). The contrast of sucient for retinal epithelial cells, but not mammary telomere lengthening in hTERT-transduced ®broblasts epithelial cells to proliferate beyond senescence (Bod- and telomere shortening in hTERT-transduced BMECs nar et al., 1998; Kiyono et al., 1998). There are also suggests that factors other than hTERT limit the discrepancies regarding hTERT mediated immortaliza- capacity of telomerase to elongate telomeres in speci®c tion of various ®broblast strains. Although BJ cell types. Telomere associated proteins, such as TRF1, ®broblasts, derived from neonatal foreskin were TIN2 and POT1, or other components of the immortalized by hTERT alone (Jiang et al., 1999), telomerase holoenzyme may be involved in this process we have shown that MRC-5 foetal lung ®broblasts (Baumann and Cech, 2001; Kim et al., 1999; van transduced with hTERT succumbed to crisis and were Steensel and de Lange, 1997). not universally immortalized (Franco et al., 2001; MacKenzie et al., 2000). Malignant transformation of BMECs The present investigations and several previous studies have demonstrated that SV40T-transfected Other than eects on replicative lifespan, constitutive ECs encounter a growth crisis after an extended period expression of hTERT alone did not confer malignant of proliferation. SV40T transformed ECs may prolif- changes. Post-senescent BMhTERT cells were depen- erate for up to 1 year or 100 PDLs before crisis occurs dent on ECGF for growth, displayed a normal (Fickling et al., 1992; Gimbrone and Fareed, 1976; morphology, expressed EC-speci®c antigens and re- Hohenwarter et al., 1992). Gimbrone and Fareed tained normal EC functions. In contrast to BMhTERT (1976) have reported outgrowth of a presumably cells, BMSVTs were partially transformed. Although immortal clone from 1 out of 6 SV40T transformed BMSVTs retained normal EC morphology and grew at EC cultures that went into crisis. In the present study, a normal rate, these cells also exhibited an extended induction of hTERT enabled all four cultures of lifespan, reduced expression of certain EC antigens and BMSVTs to escape crisis without any proliferative lag grew independently of ECGF. Reduced expression of or reduction in growth rate, even though some KDR and CD34 on BMSVTs may have been a transduction eciencies were very low. At the time of consequence of extended propagation, since expression writing, hTERT-transduced BMSVTs were 245 ± 516 of these antigens was also suppressed on late passage PDLs beyond senescence and showed no sign of control BMECs. On the other hand, deregulation of growth arrest or cell death. These results con®rm that VWF and P-CAM appeared to be related to transfor- for ECs transformed with SV40T, telomerase is the mation by SV40T, as expression of these molecules was limiting factor at crisis. Our results also show that stable in late passage control BMECs and post- immortalization of BMECs with SV40T plus hTERT is senescent BMhTERT cells. Previous reports have also much more ecient than hTERT alone. described down regulation of VWF during propagation of SV40T transformed ECs (Hohenwarter et al., 1992). Expression of EC antigens was further reduced on Regulation of telomere length in BMECs BMSVThTERT cells, possibly due to the longer culture Despite very high levels of telomerase, telomeres in the periods. The in vivo signi®cance of these ®ndings maybe BMhTERT-1 cells continued to shorten as the cells re¯ected by observations of downregulation of VWF proliferated beyond senescence. Similarly, we also and P-CAM on late stage, compared with early stage observed telomere shortening in a BMEC clone that haemangiomas (Takahashi et al., 1994). was transfected with an hTERT expression plasmid SV40T conferred BMECs with anchorage indepen- and expressed telomerase at levels comparable to a dent growth. However, while BMSVTs generated neuroblastoma cell line (data not shown). Our TRF clusters and tiny colonies at a low frequency, results are consistent with the ®ndings of Yang et al. BMSVThTERTs had a high cloning eciency and (1999), who demonstrated telomeric shortening to 5 ± 7 formed much larger, denser colonies. Variations in the Kbp in control ECs at senescence, with continued cloning eciency among the four dierent telomere shortening in hTERT-transduced ECs that BMSVThTERT cell lines did not reach statistical proliferated beyond senescence. Similarly, Zhu et al. signi®cance. These relatively minor variations were (1999) demonstrated drastic telomere shortening in not surprising, considering the BMSVThTERT cells hTERT plus SV40T immortalized ®broblasts. Observa- were extensively passaged and likely to be genomically tions of telomere shortening in the presence of high unstable as a consequence of SV40T-mediated inactiva- levels of telomerase and during extended periods of tion of P53. On the other hand, the dramatic increase proliferation has led to the proposal that in addition to in cloning eciency and the larger size of colonies maintaining telomere length, telomerase may promote generated by BMSVThTERT cells compared with proliferation by means of a `capping' function (Black- BMSVTs is a signi®cant ®nding. The contrasting burn, 2000; Zhu et al., 1999). By capping telomeres, results between BMSVThTERT and BMSVT cells

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4208 cannot be singularly explained by the increased speci®cities and/or dierent micro-environmental re- proliferative potential conferred by hTERT. BMSVTs quirements (Elenbaas et al., 2001). However, it is also that were assayed 25 PDLs before crisis had sucient likely that dierent stocks of mice, the level of Ras proliferative potential to form a colony 1 mm in oncogene expression and/or the speci®c Ras alleles diameter. Growth rate also cannot account for the employed aected tumorigenicity of the cell lines dierence in colony formation, as there was no developed in the past and present studies. signi®cant dierence in the rate of proliferation of In the present study, mutated N-ras was used to BMSVTs compared with BMSVThTERTs (data not transform BMSVThTERT cells, whereas in former shown). Immortalization with hTERT may therefore, studies H-ras was combined with SV40T and hTERT either directly or indirectly, result in additional proper- to transform human ®broblasts and epithelial cells ties that promote colony formation. For instance, (Elenbaas et al., 2001; Hahn et al., 1999). H-ras and N- deregulated expression of adhesions molecules, as ras were previously shown to have distinct cell type observed on BMSVThTERT cells, may be favourable transformation speci®cities in vitro (Maher et al., 1995). for anchorage-independent growth. Resistance to cell The cell type speci®city of dierent Ras alleles is also death is also likely to promote colony formation. re¯ected by the segregation of speci®c mutations In contrast to the high cloning eciency of among dierent human cancers. For instance, while BMSVThTERT cells, recent investigations indicated N-ras and K-ras mutations are frequent in haemato- that co-expression of SV40T and hTERT was not logic malignancies, H-ras is often mutated in epithelial sucient for anchorage independent growth of human tumours (Bos, 1989). A clinically relevant animal ®broblasts and epithelial cells. For the latter cell types, model of liver angiosarcoma implicated N-ras and K- co-expression of oncogenic Ras, in addition to SV40T ras, but not H-ras, in malignant transformation of ECs and hTERT was necessary for in vitro colony (Froment et al., 1994). Our study con®rms the potency formation (Elenbaas et al., 2001; Hahn et al., 1999). of N-ras in transformation of human ECs. However, it Thus it appears that the requirements for anchorage is also possible that K-ras and H-ras will be independent growth may be cell-type speci®c. It may be transforming in this model. signi®cant that ECs are ontologically related to In a previous study of transformation human ECs, haematopoietic cells, which are highly clonogenic human umbilical vein ECs (HUVECs) were transfected regardless of transformation events. In our investiga- with Human Papilloma Virus E6 plus E7 genes (HPV tions, cells that co-expressed Ras produced colonies E6/E7) (Rhim et al., 1998). Transfection with HPV E6/ with an altered morphology. Changes in colony E7 led to outgrowth of an immortal clone, which had morphology may have been due to increased motility spontaneously upregulated telomerase. The immortal conferred by Ras (Fox et al., 1994). clone was subsequently infected with replication BMSVThTERT cells formed small transient tumours, competent retrovirus encoding v-K-ras. Since both which may be analogous to the transient tumours HPV E6/E7 and SV40T inactivate P53 and RB tumour generated by human ECs transfected with Polyoma suppressor pathways, our study and the investigation middle T antigen (Primo et al., 2000). The latter by Rhim et al. (1998) clearly implicate a common set of investigation compared transient EC tumour growth molecular events in the tumorigenic conversion of with human haemangiomas, which are benign and human ECs. Nevertheless, it cannot be concluded that undergo spontaneous regression. In the present study, abrogation of P53 plus RB, combined with activation BMECs that expressed N-ras in addition to SV40T and of telomerase and expression of oncogenic Ras is hTERT formed large, rapidly growing, metastatic sucient for malignant transformation of human ECs. tumours that did not regress. These results con®rm Both HPV E6/E7 and SV40T may have additional and extend investigations that demonstrated tumori- transforming properties that complement inactivation genic conversion of human ®broblasts and epithelial of P53 and RB. Also, it cannot be ruled out that helper cells through co-expression of hTERT, SV40T and virus contributed to transformation of HUVECs in the oncogenic H-ras (Elenbaas et al., 2001; Hahn et al., study by Rhim et al. (1998). 1999). However, some minor dierences in tumorigeni- In conclusion, the current investigations provide a city have been observed among the dierent cell types. detailed multi-stage model of malignant transformation For instance, tumour formation by mammary epithelial of human ECs. Sequential introduction of hTERT, cells occurred after a longer latency and with a lower SV40T and oncogenic N-ras, resulted in gradual loss of incidence than the tumours generated by transformed EC function and co-incidental acquisition of a BMECs and ®broblasts (Elenbaas et al., 2001). Also, malignant phenotype. Our results show some devia- the transformed BMECs were metastatic, whereas no tions from previous investigations where the same metastases were detected in mice injected with human genes were introduced (either singularly or together) epithelial or ®broblast tumour cells (Elenbaas et al., into alternative human cell types. On the other hand, a 2001; Hahn et al., 1999). Despite an overtly transformed combination of genes analogous to the set that phenotype, BMEC tumour cells formed capillary-like transformed human ®broblasts and epithelial cells also structures in vivo, indicating that the malignant BMECs transformed BMECs to a fully malignant phenotype. retained some EC functions. The subtle dierences in These investigations provide a useful model for further tumorigenicity of dierent cell types transfected with studies of EC biology, particularly with regard to hTERT, SV40T and oncogenic Ras may re¯ect cell type angiogenesis and malignant transformation. In addi-

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4209 tion, this model oers opportunity for testing potential expresses the ecotropic receptor for M-MLV (AdEcoR) therapeutics designed to curtail aberrant EC growth. (Scott-Taylor et al., 1998). For transduction with AdEcoR, BMSVTs were initially seeded at 26104 cells per well in a 6- well plate. The following day, the cells were washed with serum-free QBSF60 media (Quality Biologicals Inc., Gaithers- burg, MD, USA), then incubated with AdEcoR at a Materials and methods multiplicity of infection of 50 in QBSF60 for 4 h at 378C/ 5% CO . Two days after AdEcoR infection, the cells were Cell culture 2 infected with either LN-ras2 or ecotropic MGFP by BMECs were isolated from the iliac crests of normal donors incubation in undiluted viral stocks supplemented with as described (Ra®i et al., 1994). BMECs were cultured in polybrene. ECGM, except where indicated. Cultures of ECs PBEC were established by culturing mononuclear cells (MNCs) in Protein extraction and telomerase expression ECGM (Peichev, 2000). MNCs were obtained from leuka- pharesis products of cancer patients who were treated with Telomerase expression was quanti®ed using the PCR-based granulocyte colony stimulating factor and cyclophosphamide telomeric repeat ampli®cation protocol as described pre- for progenitor cell mobilization. ECGM is Medium 199 viously (Kim et al., 1994). Aliquots of 2 mg of protein were (Biowhittaker, Walkersville, MD, USA) supplemented with assayed in 25 ml reactions, which included a 32P-g-ATP end- 20% foetal bovine serum (FBS, Hyclone, Logan, UT, USA), labelled TS primer and 0.01 amol of an internal PCR control 3% human serum, 20 mg/ml EC growth factor (Biomedical (TSNT). The TSNT PCR product is 36 bp and runs 14 bp Technologies, Stroughton, MA, USA), 90 mg/ml sodium below the smallest size-fractionated TRAP-derived species. heparin (Sigma, St. Louis, MO, USA), 2 mM L-glutamine, Each TRAP assay also included a negative control (H20), 80 U/ml penicillin and 80 mg/ml streptomycin. Where in- positive control (protein extract from the SK-N-SH neuro- dicated, BMECs were grown in an alternative complete EC blastoma cell line) and the R8 quanti®cation oligonucleotide. medium, referred to as EBM2 (EGM-2-MV Bulletkit system, PCR products were fractionated through a 12.5% poly- Biowhittaker, Walkersville, MD, USA). BMSVTs (previously acrylamide gel and visualized by phosphorimaging. Telomer- referred to as BMEC-1) are human BMECs that were ase activity, or total product generated (TPG) was calculated transformed with SV40T by transfection with the plasmid as a fraction of TPG in the neuroblastoma control: pRSVT (Candal et al., 1996). pRSVT encodes both the SV40 TPG=1006[(T ± B)/CT]/[(NB ± B)/CNB], where abbrevia- large T and small T antigens. BMSVTs grow independently tions are as follows; T is the radioactive signal in test extract; of VEGF and for the present investigations, were maintained B is background activity in the negative control, CT is in Iscove's Modi®ed Dulbecco's Medium (IMDM, Gibco activity of the TSNT band in the test sample, NB is the signal BRL Life Technologies, Grand Island, NY, USA) with 20% from neuroblastoma cell line and CNB is the activity of the FBS. HT1080 cells were cultured in Dulbecco's Modi®ed TSNT band in the neuroblastoma control. Eagles medium (DMEM, Gibco BRL Life Technologies) plus 10% FBS. All cells were grown at 378C/5% CO . 2 Reverse transcriptase-polymerase chain reaction (RT ± PCR) For analysis of N-ras RNA expression, RT ± PCR was Retroviral infections performed on cell lysates. Two6105 cells were collected in All retroviral vectors employed in these studies were microcentrifuge tubes, washed in phosphate buered saline replication defective and derived from Moloney murine (PBS), then resuspended in 50 ml 0.1% diethylpyrocarbonate M-MLV. Construction of the vectors MGFP and MTIG (Sigma) and snap frozen on dry ice. Frozen cell pellets were was previously described (MacKenzie et al., 2000). MGFP is thawed by heating to 908C for 10 min, then cooled to 48C a single gene vector that encodes GFP only. MTIG is a and clari®ed by centrifugation at 13 000 g for 5 min at 48C. bicistronic vector with the hTERT cDNA linked to an Lysates were cleared of contaminating DNA by incubating internal ribosomal entry site and GFP. MGFP and MTIG 20 ml of cell lysate with 20 Units of RNAse Inhibitor vectors were packaged into viral envelopes derived from (Boehringer Mannheim, Indianapolis, IN, USA) and gibbon ape leukaemia virus or the vesicular stomatitis virus 0.5 Units of RQ1 RNAse-free DNAse (Promega, Madison, G glycoprotein. The LN-ras2 vector is a single gene vector WI, USA) in a 25 ml reaction of 16PCR buer (Perkin that encodes human N-ras with an activating point mutation Elmer), for 30 min at 378C. For cDNA synthesis, a standard at codon 12 (MacKenzie et al., 1999). Viral stocks of the LN- reverse transcription reaction was performed with 25 ml ras2 vector were collected from stably transfected cLN-ras2 DNAse-treated lysate, 10 mg of random hexamer (Pharmacia producer cells, which generate ecotropic virus. Biotech, Peapack, NJ, USA), 50 mM dNTPs (Perkin Elmer), For transductions with MGFP and MTIG, primary ECs 40 Units of RNAse Inhibitor and 200 Units of Superscript II were seeded into 6-well plates at a density of 76104 cells per reverse transcriptase (Gibco BRL) in 50 mlof16PCR buer. well 1 day prior to the ®rst round of infection. Three to ®ve PCR reactions were performed using 10 ml of cDNA, 50 mM rounds of infection were performed by incubating ECs in dNTPs, 0.5 mM of each reverse and forward primer, viral supernatant diluted 1 : 1 in ECGM with 6 mg/ml 2.5 Units of Taq polymerase (Perkin Elmer) in 50 ml polybrene (Sigma). Transduction of BMSVTs with MTIG 16PCR reaction buer. Thermal cycling conditions for the and MGFP was performed by seeding at 56104 cells per well PCR reactions for both the LN-ras2 vector and b-actin were; in 6-well plates the day before infection. Three to ®ve rounds an initial 958C for 5 min, then 35 cycles of 948C for 30 s, of infection were performed using undiluted viral supernatant 608C for 1 min and 728C for 2 min followed by a ®nal supplemented with 6 mg/ml of polybrene. Transduced cells elongation step at 728C for 10 min. The primers for were enumerated by ¯ow cytometric analysis for GFP ampli®cation of LN-ras2 vector cDNA span the 3' region expression using a Becton-Dickinson FACScan. of the N-ras cDNA and the 3' region of the vector backbone. Infection of BMSVTs with ecotropic LN-ras2 was per- The sequence of the forward LN-ras2 primer is formed following transduction with an adenoviral vector that CGAAGGCTTCCTCTGTGTAT. The reverse LN-ras2 pri-

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4210 mer is GGACCACTGATATCCTGTCT. The forward and MA, USA) and visualized by incubation with Diaminoben- reverse primers for b-actin ampli®cation are GTGGGGC- zadine (Roche Mannheim, Germany) according to the GCCCCAGGCACCA and CTCCTTAATGTCACGCAC- manufacturer's instructions. GATTTC respectively. Colony assay TRF assay and Southern blot analysis For assessing colony formation in agarose, control BMECs Genomic DNA was extracted by standard phenol/chloroform and BMECs transduced with MTIG were plated in 0.33% procedure and precipitated with 0.3 M sodium acetate and 2 agarose/20% FBS/ECGM. BMSVTs, and all cells derived volumes of ethanol. TRF assays were performed as from BMSVTs, were plated in 0.33% agarose/20% FBS/ previously described (Engelhardt et al., 1997). In brief, 5 ± IMDM. The 0.33% agarose suspension was added to a 10 mg of DNA was digested with MspI and RsaI (Boehringer preformed layer of 0.5% agarose and incubated at 378C/5% Mannheim) for 16 h at 378C, followed by electrophoresis CO2 for 14 days. through a 0.5% agarose gel. Gels were then depurinated, denatured, neutralized and transferred to Nytran nylon Tumorigenicity membranes (Schleicher and Schuell, Keene, NH, USA) using Southern blot technique. Membranes were prehybridized in Tumorigenicity was assessed by both SC and IV inoculation 66SSPE/0.5% SDS/56Denhardt's/20 mg/ml tRNA for 3 h of immunocompromised mice with ECs or tumour cells. For at 558C, then hybridized in 66SSPE/1% SDS/56Denhardt's SC injections, 5 ± 106106 cells suspended in serum-free with a 5' 32P-g-ATP end-labelled telomeric oligonucleotide IMDM were injected into the hind ¯anks of Nude mice or probe (TTAGGG)4 for 16 h at 558C. Membranes were sublethally irradiated (3.5g) NOD/SCID mice. Alternatively, washed three times in 66SSPE/0.1% SDS at room sublethally irradiated NOD/SCID mice were injected via the temperature and once at 558C for 60 s. Finally, the tail vein with 5 ± 106106 cells in serum-free IMDM. Mice membranes were rinsed in 56SSPE and exposed to were monitored for tumour formation, weight loss, and phosphorimaging plates overnight. Mean and peak TRFs general malaise. were analysed as previously reported (Engelhardt et al., 1997). Flow cytometry Quantitation of speci®c cell surface antigens was performed Uptake of Dil-Ac-LDL by staining 26105 cells with phycoerytherin or ¯uorescein ECs were incubated with 2 ml/ml Dil-Ac-LDL (Intracel, isothiocyanate-conjugated monoclonal antibodies. The cells Issaquah, WA, USA) in serum-free media at 378C for 8 h. were ®rst washed with 2% FBS/PBS, then incubated in The cells were then washed and checked for Dil-Ac-LDL conjugated antibodies diluted in 2% FBS/PBS for 30 min at uptake by ¯uorescence microscopy. Dil-Ac-LDL uptake was 48C. Flow cytometric analysis was performed on a Becton- quanti®ed by FACS analysis. Dickinson FACScan.

Immunochemical staining For detection of VWF, ECs were grown on 8-well chamber slides that were coated with 0.1% gelatin/PBS. When semi- con¯uent, the cells were rinsed with PBS and ®xed in ice cold methanol for 6 min. After ®xation, endogenous peroxidase Acknowledgements was blocked by incubation in 0.1% H202 for 30 min at room This work was supported by NCI grant CA59350, NHLBI temperature. The cells were then incubated in 2% goat serum grant HL 61401 and the Gar Reichman Fund of Cancer (GS)/PBS for 40 min at room temperature. Following a Research Institute. We thank Ms Dianna Ngok, Mr Jason 5 min rinse in PBS, the cells were incubated in 5 mg/ml anti- Anselmo, Mr Sheik Baksh and Mr Harry Satterwaite for VWF monoclonal anti-body (Pharmingen, San Diego, CA, technical assistance, Mr George Nam for preparation of USA) in 2% GS/PBS overnight at 48C. Bound antibody was ECGM and EC isolation. We thank Geron Corporation detected with 2nd step reagents from a Unitect Immunohis- for the hTERT cDNA and Imclone Systems Incorporated tochemistry System (Oncogene Research Products, Boston, for antibodies to VEGF receptors.

References

Andrejauskas E and Moroni C. (1989). EMBO J., 8, 2575 ± Couch V, Lindor NM, Karnes PS and Michels V. (2000). 2581. Mayo Clin. Proc., 75, 265 ± 272. Baumann P and Cech TR. (2001). Science,292, 1171 ± 1175. Counter CM, Avilion AA, LeFeuvre CE, Stewart NG, Bergers G, Javaherian K, Lo K, Folkman J and Hanahan D. Greider CW, Harley CB and Bacchetti S. (1992). EMBO (1999). Science, 284, 808 ± 812. J., 11, 1921 ± 1929. Blackburn EH. (2000). Nature, 408, 53 ± 55. Counter CM, Hahn WC, Wei W, Dickinson Caddle S, Bodnar AG, Ouellette M, Frolkis M, Holt SE, Chiu C-P, Beijersbergen RL, Lansdorp PM, Sedivy JM and Wein- Morin GB, Harley CB, Shay JW, Lichtsteiner S and berg RA. (1998). Proc. Natl. Acad. Sci. USA, 95, 14723 ± Wright WE. (1998). Science, 279, 349 ± 352. 14728. Bos JL. (1989). Cancer Res., 49, 4682 ± 4689. Dickinson MA, Hahn WC, Ino Y, Ronford V, Wu JY, CandalFJ,Ra®iS,ParkerJT,AdesEW,FerrisB,Nachman Weinberg RA, Louis DN, Li FP and Rheinwald JG. RL and Keller KL. (1996). Microvasc. Res., 52, 221 ± 234. (2000). Mol. Cell Biol., 20, 1436 ± 1447.

Oncogene Malignant transformation of human endothelial cells KL MacKenzie et al 4211 Drolet BA, Esterly NB and Frieden IJ. (1999). New England MarkRJ,PoenJC,TranLM,FuYSandJuillardGF. J. Med., 341, 173 ± 181. (1996). Cancer, 77, 2400 ± 2406. Elenbaas B, Spirio L, Koerner F, Fleming MD, Zimonjic Moyzis RK, Buckingham JM, Cram LS, Dani M, Deaven DB, Donaher JL, Popescu NC, Hahn WC and Weinberg LL, Jones MD, Meyne J, Ratli RL and Wu J-R. (1988). RA. (2001). Genes Dev., 15, 50 ± 65. Proc. Natl. Acad. Sci. USA, 85, 6622 ± 6626. Engelhardt M, Kumar R, Albanell J, Pettengell R, Han W O'HareMJ,BondJ,ClarkeC,TakeuchiY,AthertonAJ, and Moore MAS. (1997). Blood, 90, 182 ± 193. Berry C, Moody J, Silver ARJ, Davies DC, Alsop AE, Fearon ER and Vogelstein B. (1990). Cell, 61, 759 ± 767. Neville AM and Jat PS. (2001). Proc. Natl. Acad. Sci. Fickling SA, Tooze JA and Whitley GSJ. (1992). Exp. Cell. USA, 98, 646 ± 651. Res., 201, 517 ± 521. Peichev M, Naiyer AJ, Pereira D, Zhu Z, Lane WJ, Williams Fox PL, Gaurisankar S, Dobrowolski SF and Stacey DW. M, Oz MC, Hicklin DJ, Witte L, Moore MAS and Ra®i S. (1994). Oncogene, 9, 3519 ± 3526. (2000). Blood, 95, 952 ± 958. Franco S, MacKenzie KL, Dias S, Alvarez S, Ra®i S and Primo L, Roca C, Ferrandi C, Lanfrancone L and Bussolino Moore MAS. (2001). Exp. Cell Res., 268, 14 ± 25. F. (2000). Oncogene, 19, 3632 ± 3641. FromentO,BoivinS,BarbinA,BancelB,TrepoCand Pruitt K and Der CJ. (2001). Cancer Letters, 171, 1±10. Marion MJ. (1994). Cancer Res., 54, 5340 ± 5345. Ra®i S, Shapiro F, Rimarachin J, Nachman RL, Ferris B, Gimbrone Jr MA and Fareed GC. (1976). Cell, 9, 685 ± 693. Weksler B, Moore MAS and Asch AS. (1994). Blood, 84, Girardi AJ, Jensen FC and Koprowski H. (1965). J. Cell. 10 ± 19. Comp. Physiol., 65, 69 ± 84. Rak J, Mitshuhashi Y, Bayko L, Filmus J, Shirasawa S, Hahn WC, Counter CM, Lundberg AS, Beijersbergen RL, Sasazuki T and Kerbel RS. (1995). Cancer Res., 55, 4575 ± Brooks MW and Weinberg RA. (1999). Nature, 400, 464 ± 4580. 468. Ramirez RD, Morales CP, Herbert B-S, Rohde JM, Passons Halvorsen TL, Leibowitz G and Levine F. (1999). Mol. Cell. C, Shay JW and Wright WE. (2001). Genes Dev., 15, 398 ± Biol., 19, 1864 ± 1870. 403. Harley CB, Futcher AB and Greider CW. (1990). Nature, RhimJS,TsaiWP,ChenZQ,VanWaesC,BurgerAMand 345, 458 ± 460. Lautenberger JA. (1998). Carcinogenesis, 19, 673 ± 681. Hay¯ick L and Moorhead PS. (1961). Exp. Cell. Res., 25, Scott-Taylor TH, Gallardo HF, Gansbacher B and Sadelain 585 ± 621. M. (1998). Gene Ther., 5, 621 ± 629. HohenwarterO,ZinserE,SchmatzC,RukerFandKatinger Seeburg PH, Colby WW, Capon DJ, Goeddel DV and H. (1992). J. Biotech., 25, 349 ± 356. Levinson AD. (1984). Nature, 312, 71 ± 73. Jiang XR, Jimenez G, Chang E, Frolkis M, Kusler B, Sage Shay JW, Pereira-Smith OM and Wright WE. (1991). Exp. M,BeecheM,BodnarAG,WahlGM,TlstyTDandChiu Cell Res., 196, 33 ± 39. CP. (1999). Nat. Genet., 21, 111 ± 114. Shay JW, Van Der Haegen BA, Ying Y and Wright WE. Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, (1993). Exp. Cell. Res., 209, 45 ± 52. Ho PLC, Coviello GM, Wright WE, Weinrich SL and Smith J and Pereira-Smith OM. (1996). Science, 273, 63 ± 66. Shay JW. (1994). Science, 266, 2011 ± 2014. Takahashi K, Mulliken JB, Kozakewich HPW, Rogers RA, Kim S, Kaminker P and Campisi J. (1999). Nat. Genet., 23, Folkman J and Ezekowitz RAB. (1994). J. Clin. Invest., 405 ± 312. 93, 2357 ± 2364. Kiyono T, Foster SA, Koop JI, McDougall JK, Galloway van Steensel B and de Lange T. (1997). Nature, 385, 740 ± DA and Klingelhutz AJ. (1998). Nature, 396, 84 ± 88. 743. Klingelhutz AJ, Barber SA, Smith PP, Dyer K and Vogelstein B, Fearon ER, Hamilton SR, Kern SE, Preisinger McDougall JK. (1994). Mol. Cell. Biol., 14, 961 ± 969. AC, Leppert M, Nakamura Y, White R, Smits AM and Land H, Parada LF and Weinberg RA. (1983). Science, 222, Bos JL. (1988). New Eng. J. Med., 319, 525 ± 532. 769 ± 770. YangJ,ChangE,CherryAM,BangsCD,OeiY,BodnarA, MacKenzie KL, Dolnikov A, Millington M, Shounan Y and BronsteinA,ChiuC-PandHerronGS.(1999).J. Biol. Symonds G. (1999). Blood, 93, 2043 ± 2056. Chem., 274, 26141 ± 26148. MacKenzie KL, Franco S, May C, Sadelain M and Moore Zhu J, Wang H, Bishop JM and Blackburn EH. (1999). Proc. MAS. (2000). Exp. Cell. Res., 259, 336 ± 350. Natl. Acad. Sci. USA, 96, 3723 ± 3728. Maher J, Baker DA, Manning M, Dibb NJ and Roberts IA. (1995). Oncogene, 11, 1639 ± 1647.

Oncogene