Telomere shortening in primary human fibroblasts results in replicative senescence, which can be overcome by telomerase (hTERT) overexpression. However, because immortalization is one of the hallmarks of malignant transformation, careful analysis of hTERT-immortalized cells is of crucial importance for understanding both processes. To this end, we infected WI-38 fibroblasts with a retrovirus carrying the hTERT cDNA and analyzed their proliferative behavior during 600 days [∼500 population doublings (PDLs)] of continuous culture. Growth of three independent mass cultures was uniform for ∼150 PDLs after telomerase infection, followed by a progressive acceleration of growth in two of three cultures. Expression of p16INK4A was significantly elevated in the immortalized cells but gradually disappeared during the accelerated growth phase. This alteration correlated with loss of the contact inhibition response and conferred the cells with sensitivity to H-Ras-induced transformation. In contrast, the p53- and pRb-mediated checkpoints such as the DNA damage response, chromosomal stability and entry into quiescence remained intact, irrespective of INK4A locus expression. Importantly, detailed examination of one of the WI-38/hTERT cultures during the accelerated growth phase revealed overexpression of the c-myc and Bmi-1 oncogenes, as well as loss of p14ARF expression. Collectively, our results indicate that although hTERT-immortalized cells behave similarly to primary cells during the first 150 PDLs, long-term growth in culture may favor the appearance of clones carrying potentially malignant alterations.
HDFs, 4 when passaged in culture, gradually reduce their proliferation rate and enter an irreversible growth arrest termed replicative senescence (1) . Senescent fibroblasts are characterized by flattened morphology, enlarged cell size, diminished DNA replication, positive β-galactosidase staining at neutral pH (SA-β-GAL), and exhibit a distinct pattern of gene expression (2) . An additional feature, which is correlated with the passaging of human primary cells in vitro, is a progressive shortening of telomeres because of an end-replication problem (3) . Telomeres consist of repetitive TTAGGG sequences that in complex with telomere associated proteins, create special T-loop structures at the ends of each chromosome (4) . Both telomeric DNA and trans-acting protein factors protect the chromosome ends from recombination, fusion, or degradation, and finally, prevent the chromosome ends from being recognized as damaged DNA (5 , 6) . In the absence of a specific repair mechanism, primary somatic cells stop proliferating when a critical telomere length is reached. The initiation of this cell cycle arrest is mainly attributed to wild-type p53 tumor suppressor activation. In addition, a role for p16INK4A protein in this process was recently suggested (7, 8, 9, 10) .
In contrast to normal cells, the majority of tumor-derived cells express telomerase (hTERT), an RNA dependent DNA polymerase, which is able to catalyze telomere elongation (11 , 12) . Several studies demonstrated that overexpression of hTERT in HDFs and several other cell types prevents telomere shortening and is sufficient to bypass replicative senescence and immortalize these cells (13, 14, 15, 16) . The above observations have led to the telomere-dependent theory of senescence. However, the contribution of telomere shortening to human aging is still a controversial issue (17 , 18) .
The induction of replicative senescence, as well as the maintenance of a nonproliferative state, requires proper functioning of the p53- and pRb-controlled signaling pathways. The role of p53 in senescence is mainly attributed to its ability to transactivate the p21WAF1 cyclin dependent kinase inhibitor, which, in turn, is sufficient to terminate cell cycle progression. Induction of p16INK4A and subsequent inactivation of cyclinD/CDK4/6 complexes provides an additional mechanism that acts in replicative senescence (8 , 9 , 19 , 20) . Indeed, inactivation of these tumor suppressor genes by viral oncogenes such as E6 and E7 of human papillomavirus type 16 virus or large T antigen of SV40 virus allows cells to escape the short telomere associated checkpoint and proliferate for an additional 20–40 PDLs until the cells fall into crisis (21, 22, 23, 24, 25) . Rare immortal clones that recover after the crisis ultimately reactivate telomerase or the alternative lengthening of telomeres pathway (26 , 27) . Tight involvement of tumor suppressors such as p53, pRb, and p16INK4A in the induction of senescence in vitro on the one hand and their inactivation in malignant transformation on the other indicate that replicative senescence may serve as a tumor protective barrier in vivo, as well. Although hTERT overexpression in primary cells was not found to be associated with a malignant phenotype (28 , 29) , its ability to immortalize cells allows the expansion of a cell population far beyond the limits imposed by replicative senescence. Theoretically, three scenarios are possible: (a) hTERT-expressing cells finally cease proliferation because of factors independent of telomere length; (b) hTERT-expressing cells continue to proliferate but do not acquire malignancy-associated changes; and (c) continued proliferation of hTERT-immortalized cells may select for alterations that confer them with additional growth advantages. Taking into consideration the appeal of using hTERT in cell-based therapies and the intimate association of telomerase activity with malignancies, a greater understanding of the effects of stable hTERT overexpression is important. To address this issue, we transduced the primary human diploid fibroblast strain WI-38 with hTERT and monitored the behavior of the immortalized cells for >600 days of continuous culturing. We observed gradual growth acceleration after ∼150 PDLs. Furthermore, expression of the p16INK4A and p14ARF genes, which was up-regulated after the hTERT-immortalized cells bypassed the predetermined replicative limit, gradually underwent silencing. We show here that inactivation of the INK4A locus was associated with a loss of the contact inhibition checkpoint and conferred the cells with sensitivity to H-Ras-mediated transformation.
MATERIALS AND METHODS
Primary human embryonic lung fibroblasts (WI-38), amphotropic, and ecotropic Phoenix retrovirus-producing cells were purchased from the American Type Culture Collection. WI-38 cells were grown in MEM supplemented with 10% FCS, 1 mm sodium pyruvate, 2 mm l-glutamine, and antibiotics. Phoenix cells were grown in DMEM supplemented with 10% FCS, 2 mm l-glutamine, and antibiotics. All of the cells were maintained in a humidified incubator at 37°C and 5% CO2. Cells were split close to confluence by incubation with trypsin and replated into a new plate at cell density of 1500 cells/cm2. PDLs were calculated using the formula: PDLs = log (cell output/cell input)/log2.
pBabe-hTERT-puro was kindly provided by Dr. Jerry Shay (University of Texas Southwestern Medical Center), pBabe-H-Ras V12-hygro and ecotropic receptor retroviral constructs were kindly provided by Dr. Doron Ginsberg (Weizmann Institute), and PLXSN-GSE56-Neo was obtained by the subcloning of GSE56 BamHI fragment from pBabe-GSE56-puro (30) into PLXSN.
Amphotropic and ecotropic Phoenix-packaging cells were transfected with 10 μg of DNA of the appropriate retroviral construct by a standard calcium phosphate procedure. Culture supernatants were collected 36–48 h after transfection and filtered. WI-38 cells were infected with the filtered viral supernatants in the presence of 4 μg/ml polybrene (Sigma) for 12 h, after which, the medium was changed. Fresh viral suspensions were added after a 24-h interval for an additional 12 h. Infected cells were selected with 1 μg/ml puromycin (5 days), 400 μg/ml G418 (14 days), or 300 μg/ml hygromycin (5 days).
Western Blotting Analysis.
Cells were lysed in Tris Triton Lysis Buffer (TLB) buffer (50 mm Tris-Cl, 100 mm NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, and 0.1% SDS) supplemented with protease inhibitor mixture (Roche) and phosphatase inhibitor mixtures I and II (Sigma) for 30 min on ice. Extracts were analyzed for protein concentration by Bradford assay. For electrophoresis, 50 μg of protein extract were dissolved in sample buffer [140 mm Tris (pH 6.8), 22.4% glycerol, 6% SDS, 10% β-mercaptoethanol, and 0.02% bromphenol blue] boiled and loaded on 10–12.5% polyacrylamide gels containing SDS. Proteins were transferred to nitrocellulose membranes. The following primary antibodies were used: mouse monoclonal anti-p53 (DO-1 and 1801; kindly provided by Dr. David Lane [Ninewells Hospital and Medical School (Dundee, Scotland)]; rabbit polyclonal anti-p53 (produced in our laboratory); mouse monoclonal anti-MDM2 [4B2, 2A9, and 2A10; kindly provided by Dr. Moshe Oren (Weizmann Institute of Science)]; rabbit polyclonal anti-p21 (C-19; Santa Cruz Biotechnology); rabbit polyclonal anti-p16 (C-20; Santa Cruz Biotechnology); mouse monoclonal anti-Ras (C-18, BD Transduction Laboratories); mouse monoclonal anti-tubulin (Sigma); and mouse monoclonal anti-Bmi-1 (229F6; Upstate Biotechnology). The protein-antibody complexes were detected using horseradish peroxidase-conjugated secondary antibodies and the Super-signal enhanced chemiluminescence system (Pierce).
Measurement of Telomere Length by the TRF Assay.
Genomic DNA was extracted by GenElute Mammalian Genomic DNA Kit (Sigma) according to the manufacturer’s recommendations. Next, 2 μg of genomic DNA were reacted according to Telo TAGGG Telomere Length Assay kit (Roche Molecular Biochemicals). Washed membranes were exposed to phosphorimaging plates for 5–60 min. The mean TRF length was determined using MacBas 2500 software according to the following formula: mean TRF = Σ (Ai)/Σ(Ai/Li), where Ai is the chemiluminiscent signal and Li is the length of the TRF fragment at position i.
Telomerase activity determinations were performed using a commercial TRAPeze kit (Intergene) according to manufacturer’s nonradioactive protocol. The cycling conditions were modified as follows: 30°C for 30 min, 94°C for 3 min; and 29 cycles of amplification: 94°C for 30 s, 56°C for 30 s, and 72°C for 30 s. Unless indicated otherwise, 500 ng lysate/telomeric repeat amplification protocol reaction were used.
Total RNA was isolated by Tri Reagent (Molecular Research Center), and 1 μg was reverse transcribed with EZ-First Strand cDNA Synthesis kit (Biological Industries, Beit Haemeck, Israel) according to the manufacturer’s protocol. Hot start PCR was carried out for 19 (GAPDH), 25 (p21WAF1), 26 (c-myc), 29 (p16INK4A, p14ARF), 35 (hTERT), and 26 (Bmi-1) cycles. The linear range of amplification was determined by varying the number of PCR cycles for each cDNA and set of primers. The primers were as follows: GAPDH, 5′-TCCACCACCCTGTTGCTGTA and 3′-ACCACAGTCCATGCCATCAC; p21WAF1, 5′-CGCGACTGTGATGCGCTAATG and 3′-GGAACCTCTCATTCAACCGCC; c-myc, 5′-CTACGTTGCGGTCACACCC and 3′-GAGGGGTCGATGCACTCTG; p16INK4A, 5′-GAGCAGCATGGAGCCTTCGGand 3′-CATGGTTACTGCCTCTGGTG; p14ARF, 5′-GAAGATGGTGCGCAGGTTCT and 3′-CCTCAGCCAGGTCCACGGG; hTERT, 5′-GCCTGAGCTGTACTTTGTCAA and 3′-CGCAAACAGCTTGTTCTCCATGTC; and Bmi-1, 5′ ACAGCCCAGCAGGAGGTATTC and 3′-GCCCAATGCTTATGTCCACTG. PCR products were separated on agarose gels and visualized by ethidium bromide staining.
SA-β-GAL Activity Staining.
Cells were washed in PBS and fixed with 2% formaldehyde/0.2% gluteraldehyde in PBS for 5 min at room temperature. Plates were stained for SA-β-gal activity, as described previously (31) .
Spectral Karyotype Analysis.
Exponentially growing cells were incubated with Colcemid (0.1 μg/ml) for 3 h, trypsinized, lysed with hypotonic buffer, and fixed in glacial acetic acid/methanol (1:3). The chromosomes were simultaneously hybridized with 24 combinatorially labeled chromosome painting probes and analyzed using the SD200 spectral bioimaging system (Applied Spectral Imaging Ltd., Migdal Haemek, Israel).
Anchorage-Independent Growth (Agar Colony Assay).
Between 1 and 2 × 104 cells were suspended in 1 ml of 2× MEM supplemented with 20% FCS, 2 mm sodium pyruvate, 4 mm l-glutamine, and mixed with 1 ml of 0.22% molten agarose, held at 60°C. The mixture was then layered on top of 1 ml of solidified 0.5% agarose in MEM supplemented with 10% FCS, 1 mm sodium pyruvate, and 2 mm l-glutamine in a 35-mm plate. The cells were incubated at 37°C and fed with fresh 0.2% agarose/MEM/FCS every 7 days. Colonies were counted after 21 days.
WI-38 cells and WI-38/hTERT from the indicated passages were trypsinized and 300-1000 cells/100-mm dish were plated out in duplicate. After 16–21 days, dishes were fixed, stained with crystal violet, and colonies were counted.
Cell Cycle Analysis.
Subconfluent cultures were labeled for 30 min with 10 μm BrdUrd (Sigma). Cell were detached with trypsin, fixed in 70% ethanol, and treated as follows (PBS washes between each step): 2 m HCl and 0.5% Triton X-100 for 30 min at room temperature; 0.1 m Na2B4O7 at pH 8.5; FITC-conjugated anti-BrdUrd (Becton Dickinson) diluted 1:3 in PBS/1% BSA/0.5% Tween 20 for 1 h at room temperature; and finally, 5 μg/ml propidium iodide and 0.1 mg/ml RNase A. Samples were analyzed by two-dimensional flow cytometry to detect both fluorescein and propidium iodide fluorescence using a fluorescence-activated cell sorter (Becton Dickinson). At least 10,000 cells were analyzed/sample.
Overexpression of hTERT Induces Immortalization of WI-38 Fibroblasts.
WI-38 human diploid fibroblasts are a widely used cell culture model for studying replicative senescence and transformation processes. To create a convenient system to study the changes which accompany those processes, we infected the WI-38 cells (at 40 PDLs) with a recombinant retrovirus encoding for hTERT and, in parallel, with its empty vector counterpart, pBabe-puro. After recovery from selection, the cells were serially passaged. Although the cells infected with the control virus (WI-38/puro) gradually ceased proliferating after ∼50 PDLs, the mass culture (three separate pools that were passaged separately) of hTERT-infected cells (WI-38/hTERT) continued to proliferate beyond the replicative senescence checkpoint and underwent up to 600 PDLs for cells initiated from the first pool (designated WI-38/hTERT, clone1), 200 PDLs for cells initiated from the second pool (designated WI-38/hTERT, clone2), and 175 PDLs for cells initiated from the third pool (designated WI-38/hTERT, clone3) without signs of growth retardation (Fig. 1A) ⇓ . Because we grew WI-38/hTERT, clone1 (further referred to as WI-38/hTERT) for the longest period of time, the majority of the assays presented in this study were performed on this cell population, unless otherwise indicated. As expected, WI-38/puro, while attaining senescence, adopted a flattened and enlarged morphology and ∼35% of the cells showed positive staining for SA-β-GAL. In contrast, WI-38/hTERT cells at the same time points did not exhibit significant SA-β-GAL staining (<5%; supplementary Fig. 1 ⇓ ).
As shown in Fig. 1B ⇓ , hTERT infection resulted in the appearance of telomerase activity, comparable with the activity seen in the cancer cell line, H1299. Parental WI-38 and the cells infected with the control virus did not exhibit any detectable endogenous telomerase activity. In addition, expression levels of hTERT were tested in parental and hTERT infected WI-38 cells by semiquantitative RT-PCR (Fig. 1C) ⇓ . Primary human cells express very low levels of hTERT (32) because of promoter repression, however, WI-38/hTERT express high levels of exogenous hTERT mRNA.
The shortening of telomeres to a critical length contributes to the initiation of the senescence program (3 , 7) . To test the involvement of telomeres in the in vitro aging of WI-38 fibroblasts and the effect of hTERT overexpression on the telomeres, the mean TRFL was determined as a function of progressive cell passaging (Fig. 1D) ⇓ . Although the WI-38/puro cells exhibit a gradual shortening of telomeres (∼7 kb in young cells versus ∼4 kb in old cells), the WI-38/hTERT cells elongate their telomeres continuously (from ∼5 to ∼10 kb) as a function of successive passages in culture. Interestingly, in addition to length, the pattern of telomere length distribution changes as a function of hTERT overexpression. Primary cells exhibit telomeres with heterogeneous TRFLs, whereas after hTERT expression, TRFLs become more uniform in length (Fig. 1D) ⇓ .
Thus, introduction of the hTERT catalytic subunit into WI-38 primary human fibroblasts results in a significant extension of life span, supporting the hypothesis that telomere shortening is the primary cause of replicative senescence in this cell type.
Growth Characteristics of hTERT-Expressing Cells.
To characterize the molecular changes accompanying the hTERT-induced immortalization process, we followed the proliferative behavior of WI-38/hTERT over >600 days of continuous logarithmic growth. As evident from Fig. 2A ⇓ , the period of extended life span could be divided roughly into two main stages, according to the proliferation rate of cells expressing telomerase. The first stage is characterized by a rate of proliferation comparable with primary cells and was maintained for ∼250 days, which is roughly equivalent to 150 PDLs. The second phase is characterized by a gradual acceleration in the proliferation rate ranging between 30 and 52 PDLs/50 days. We shall refer to the cells in the first growth stage as WI-38/hTERTslow and those at the second stage as WI-38/hTERTfast.
An important growth parameter that distinguishes between primary and transformed cells is the loss of contact inhibition by the latter. After the observation that a significantly higher proliferation rate was evident in the late passages of hTERT-immortalized cells, their ability to grow when the culture reached confluence was assessed (Fig. 1B) ⇓ . Surprisingly, we found that WI-38/hTERTfast (assayed at 289 PDLs) were able to grow to saturation density exceeding that of primary WI-38 cells (assayed at 25, 32, and 48 PDLs) or WI-38/hTERTslow (assayed at 52 and 92 PDLs), by as much as 4.5-fold. This newly acquired proliferation capability at high saturation density, observed in WI-38/hTERTfast, is indicative of a defect in their response to contact inhibition.
Cell growth at clonal density (CFE) often reveals the cumulative effects of stress that individual cells experience during culture, including in vitro aging (17) . These effects could be masked in mass culture experiments such as those we conducted. In addition, an intact stress response is dependent on functional p53 and/or pRb tumor suppressor pathways. To reveal the effect of hTERT-mediated immortalization on CFE, cells from different passages were seeded at clonal density (Fig. 2C) ⇓ . In agreement with early studies that made use of embryonic lung fibroblasts, the CFE of early passages of WI-38 cells was ∼10% (Fig. 2C ⇓ ; Ref. 33 ). In contrast, CFE was decreased by 10-fold among WI-38/hTERTslow (tested at 90 and 131 PDLs) as compared with primary cells. The significant inhibition of CFE in WI-38/hTERTslow during the extended phase of the life span induced by telomerase supports the existence of additional mechanisms, independent of telomere-length, that may limit the life span of primary cells. CFE of WI-38/hTERTfast (tested at 353PDLs) was significantly higher when compared with the early passages of primary and WI-38/hTERTslow cells. This indicates the selection of a population with better adaptation to extended life span in vitro. Taken together, examination of several growth parameters in WI-38/hTERT reveals a biphasic behavior of the immortalized cells. A similar pattern of growth acceleration was observed in WI-38/hTERT, clone2 after 170 PDLs. In contrast, no significant change in the proliferation was evident in WI-38/hTERT, clone3 (data not shown), which retained the slow growth phenotype. WI-38/hTERTslow cells exhibited reduced CFE and arrested their growth at a cell density similar to that of primary cells. Conversely, WI-38/hTERTfast exhibited a higher proliferation rate, a defective contact inhibition checkpoint, and had a higher CFE in comparison with primary cells or with WI-38/hTERTslow.
Expression Pattern of Endogenous Cell Cycle-Related Proteins.
The irreversible growth arrest that limits proliferation of primary human fibroblasts to 50–60 PDLs is mediated by the activation of the p53 and pRb pathways (8 , 9 , 34) . In addition, in the vast majority of human cancers these pathways are defective, further indicating the importance of these pathways in immortalization and transformation (35) . To elucidate the molecular determinants accompanying hTERT-mediated immortalization, we performed expression analysis of several well-known components of the p53 and pRb pathways. p16INK4A is a critical regulator of the pRb pathway and is known to be up-regulated in response to stress occurring during prolonged tissue culture. In some cell types such as breast epithelium and keratinocytes, p16INK4A may prevent hTERT-induced immortalization (36, 37, 38, 39) . By semiquantitative RT-PCR analysis, we found that p16INK4A expression is up-regulated at the WI-38/hTERTslow stage (tested at 56, 128, and 168 PDLs) as compared with proliferating young WI-38 and achieves its maximum expression level ∼130 PDLs, followed by a gradual decline and loss of expression (Fig. 3A) ⇓ . p14ARF is an upstream regulator of p53 stability, and it shares exons 2 and 3 with p16INK4A, although using alternative reading frames (40 , 41) . Both genes reside in the INK4Alocus. The p14ARF expression pattern, resembling that of p16INK4A, was decreased in WI-38/hTERTfast passages. To substantiate our findings regarding the silencing of INK4A locus genes, we also monitored their expression in WI-38/hTERT, clone2 and clone3. Both WI-38/hTERT clone1 and clone2 exhibited a gradual decrease in p16INK4A and p14ARF expression to almost undetectable levels at ∼200 PDLs. In contrast, expression of p16INK4A in clone3 was increased and did not show a decreasing trend at any time point tested.
The Bmi-1 protein is required for the transcriptional repression of the INK4A locus (42, 43, 44) . We therefore tested its expression pattern during hTERT immortalization. As seen in Fig. 3A ⇓ , expression of Bmi-1 was strongly up-regulated in correlation with silencing of the p16INK4A and p14ARF genes. The c-myc transcription factor demonstrated an inverse correlation with INK4A locus expression, i.e., its expression was relatively low in primary and WI-38/hTERTslow cells (tested at 56, 128, and 168 PDLs) and was strongly enhanced in the WI-38/hTERTfast (tested at 216, 295, and 339 PDLs) cells (Fig. 3A) ⇓ . In contrast to the above genes, we did not detect significant changes in the expression pattern of p21WAF1 throughout the life span of the hTERT-immortalized cells.
The results obtained at the RNA level were additionally confirmed and expanded by protein analysis using Western blotting. As shown in Fig. 3B ⇓ , in agreement with the pattern of p16INK4A mRNA expression, p16INK4A protein levels were strongly elevated during the slow proliferative phase (tested at 140 and 160 PDLs), and its expression was completely lost at later time points. Unfortunately, we were unable to detect endogenous p14ARF protein in WI-38 cells, possibly because of low levels of its expression. p53 and p21WAF1 protein levels did not show significant changes at the different stages of immortalization. c-Myc protein was almost undetectable by immunoblotting in primary WI-38 as well as in WI-38/hTERTslow cells (tested at 140 and 160 PDLs). However, its level rose dramatically in WI-38/hTERTfast cells (tested at 376 and 485 PDLs) and attained expression comparable with the cancer cell lines H1299 and HeLa (Fig. 3B) ⇓ .
Finally, we found similar levels of Bmi-1 protein in primary and WI-38/hTERT slow cells. In contrast, WI-38/hTERTfast cells expressed increasing levels of Bmi-1 with the appearance of a faster migrating band possibly representing the hypoposphorylated form (45) of this factor. In addition, Bmi-1 protein levels and their migration pattern in WI-38/hTERTfast cells were comparable with the H1299 tumor cell line (Fig. 3B) ⇓ .
De novo methylation of the p16INK4A promoter region associated with expression silencing was found in several primary tumors and established tumor cell lines. Treatment of cells with (5-AzaC), a known inhibitor of DNA methylation often reactivates promoters silenced by methylation (46 , 47) . To determine whether the loss of p16INK4A expression is attributable to promoter methylation, WI-38/hTERTfast cells (350PDLs) were exposed to 5-AzaC, and levels of the p16INK4A protein were determined. Inhibition of DNA methylation in WI-38/hTERTfast partially restored p16INK4A expression in a time-dependent manner (Fig. 3C) ⇓ . We did not detect changes in p21WAF1 or β-tubulin expression as the result of 5-AzaC treatment. The combination of high c-mycand Bmi-1 oncogene expression together with the silencing of the INK4A locus may explain the accelerated growth and the loss of contact inhibition observed in WI-38/hTERTfast as compared with WI-38/hTERTslow.
Genome Stability and Functional Activity of p53 and pRb Genes in hTERT-Immortalized Cells.
Genomic aberrations of various kinds are one of the hallmarks of transformation. It is widely believed that intact p53 function is responsible for the maintenance of genome stability (48) . Spectral karyotype analysis was performed on WI-38 (30 PDLs) and on WI-38/hTERT at 65, 230, 260, 348, and 484 PDLs to assess the integrity of the genome during prolonged and continuous passaging. Parental WI-38 as well as WI-38/hTERT cells at 65PDLs exhibit a normal diploid genome without any evidence of aneuploidy or chromosomal aberrations. However, from PDL 230, a nonreciprocal translocation der(X)t(X;17) was observed in 9 of 10 of the examined metaphases. Subsequently, the translocation was identified by fluorescent in situ hybridization to represent a gain of 17q25. It is likely that this translocation represents the only clonal chromosomal change as was identified in each of the progressive samples (analyzed at 260, 348, and 484 PDLs), as well as in H-Ras transformed WI-38/hTERT-immortalized cells (Table 1 ⇓ ; supplementary Fig. 2 ⇓ ). Furthermore, using fluorescent in situ hybridization analysis, we did not detect c-mycamplification in the samples tested, suggesting that the elevated levels of c-myc are not attributable to genome instability (Table 1) ⇓ . The frequency and type of aberrations found in the WI-38/hTERT cells are comparable with those reported in other normal and hTERT-immortalized fibroblast strains (28 , 29 , 49) . It is important to note that the passage at which this translocation was first observed coincides with gradual growth acceleration and INK4A locus silencing. Thus, we conclude that hTERT-immortalized cells maintain a stable genome even after 484 cumulative PDLs. In addition, our karyotype analysis suggests that the INK4A locus does not significantly affect genome integrity at the chromosome level, at least 200 divisions after its inactivation.
Several types of DNA damage are known to activate p53, which can terminate the proliferation of cells with unrepaired or improperly repaired DNA (48) . We assessed p53 activation by treatment of WI-38 cells and their hTERT-expressing counterparts with the DNA-damaging agent, doxorubicin. Strong p53 stabilization was evident 5 h after doxorubicin treatment and was maintained for at least 48 h. The activated p53 was able to induce the expression of its downstream targets such as p21WAF1 and MDM2 (Fig. 4A) ⇓ . A similar pattern of p53 activation and induction of its transcriptional targets was observed in parental WI-38 cells as well as in hTERT-immortalized cells [WI-38/hTERTslow (48 and 96 PDLs) and WI-38/hTERTfast (353 PDLs)]. The cell cycle response of primary and hTERT-immortalized cells to this dose of doxorubicin was characterized by an almost complete S-phase disappearance and accumulation of cells in the G2 phase of the cell cycle (Fig. 4B) ⇓ . To confirm the dependence of this response on p53, we made use of WI-38/hTERT cells stably expressing a dominant negative p53 polypeptide, GSE56 (30) . The expression of GSE56 efficiently blocked p53-dependent transactivation (data not shown). The cells with inactivated p53 did not arrest after doxorubicin treatment, indicating the participation of p53 in this process (Fig. 4B) ⇓ . Importantly, we did not observe any detectable differences in the kinetics or in the extent of p53 induction between WI-38/hTERT cells, which differed in their INK4A locus expression.
Reversible growth arrest in response to growth factor depletion (quiescence) is mediated by both pRb and p53 activation (50) . Many transformed and tumor-derived cells continue to proliferate despite the depletion of growth factors. To characterize the integrity of the quiescence response as a function of hTERTexpression and of INK4A locus status, we assayed the proportion of cells in S phase under low serum conditions. WI-38/hTERTfast (295 PDLs) exhibit a quiescence response similar to that of primary cells. p53 inactivation by transfection with a dominant negative form of p53 (GSE 56) causes a delayed entrance into quiescence (data not shown). In agreement with previous studies (28 , 29 , 49) , these results suggest that hTERT-mediated immortalization does not result in gross genomic changes at the chromosomal level; cells maintain an intact p53-dependent DNA damage response and a functional quiescence checkpoint. We did not detect a significant contribution of the INK4A locus to the above responses.
Sensitivity to H-Ras-Induced Transformation Is Correlated with the Status of INK4A Locus.
Primary human diploid fibroblasts enter irreversible growth arrest with features of senescence in response to overexpression of the oncogenic Ras protein. This arrest is mediated by the concomitant activation of p53 and p16INK4A tumor suppressors (51, 52, 53) . However, the particular contribution of each of those genes is unclear. Previous studies suggested that hTERT-immortalized fibroblasts behave in a manner indistinguishable from their mortal counterparts in response to H-Ras mediated transformation (54) . Taking into consideration that the late passages of WI-38/hTERT do not express p16INK4A and p14ARF, their response to mutant Ras may provide important clues regarding Ras-mediated transformation.
We infected parental WI-38, WI-38/hTERTslow (75 PDLs) and WI-38/hTERTfast (340 PDLs) with retroviruses encoding either H-RasV12 cDNA or an empty vector counterpart. To determine the role of p53 in the response to mutant Ras overexpression, we used a retrovirus encoding a dominant negative form of p53 (GSE56). After infection and selection, the growth kinetics, morphological, and biochemical properties of the cells were analyzed (Fig. 5) ⇓ . H-RasV12 overexpression resulted in arrest of WI-38 and WI-38/hTERTslow cells. However, WI-38/hTERTfast/H-RasV12 and WI-38/hTERTfast/H-RasV12/GSE56 resumed proliferation after a transient 3-day arrest (Fig. 5A) ⇓ . Morphologically, H-RasV12-infected WI-38 and WI-38/hTERTslow cells acquired a flattened morphology similar to that of senescent fibroblasts. Conversely, WI-38/hTERTfast/H-RasV12 cells, after a delay of 3–5 days, formed foci of vigorously proliferating cells on the background of arrested cells. WI-38/H-RasV12, as well as WI-38/hTERTslow/H-RasV12, showed diminished BrdUrd incorporation versus control virus-infected cells (Fig. 5B) ⇓ , in agreement with the growth curves. In contrast, recovered pools of WI-38/hTERTfast/H-RasV12 showed an increase in BrdUrd incorporation. Ras-arrested cells demonstrated SA-β-GAL staining, indicating that the arrest exhibited features of replicative senescence. On the other hand, the WI-38/hTERTfast/H-RasV12 cells did not show staining with this marker at levels above the controls (Fig. 5C) ⇓ .
Anchorage-independent growth is a hallmark of Ras transformation (55, 56, 57) , therefore, we tested the ability of H-RasV12-infected cells to grow in soft agar. Although the WI-38/hTERTfast cells infected with the empty vector or with dominant negative p53 failed to form colonies, WI-38/hTERTfast/H-RasV12 cells were able to grow in the soft agar. Furthermore, we observed a strong synergistic effect between p53 inactivation and mutant Ras overexpression in the soft agar colony formation assay as measured by colony number and size (Fig. 5D) ⇓ . Overexpression of H-Ras was confirmed by Western blot analysis (Fig. 5E) ⇓ . Interestingly, the highly expressed p16INK4A protein in the WI-38/hTERTslow cells was not additionally induced by H-Ras. In contrast, p16INK4A could not be detected in the WI-38/hTERTfast cells after H-RasV12 overexpression.
In addition to the described effects of mutant H-Ras overexpression and p53 inactivation on cell growth parameters, we analyzed their impact on genome stability (Table 1) ⇓ . Combination of H-Ras and p53 inhibition (WI-38/hTERT/H-Ras/GSE) resulted in extensive aneuploidy and random chromosomal translocations. Interestingly, none of the detected aberrations were identical, suggesting that they do not represent changes that became clonally expanded in the population.
Taken together, our results provide strong evidence that hTERT-immortalized cells exhibit differential sensitivity to mutant Ras-induced transformation. Although the early passages of hTERT-expressing fibroblasts were resistant to H-RasV12, late passages, which do not express p14ARF and p16INK4A, were susceptible.
To characterize the changes in genome integrity and growth control that accompany the immortalization process, we used the ability of telomerase to confer somatic cells with unlimited proliferation potential. We found that hTERT-immortalized fibroblasts gradually acquired distinct features associated with a transformed phenotype. Furthermore, dramatic changes in the expression pattern of a number of tumor suppressors and oncogenes eventually became evident, altogether suggesting an increased risk of such cells to ultimately convert into malignant cells. There are a growing number of studies, which suggest that cell immortality represents a combined phenotype involving a telomere maintenance mechanism together with changes in certain growth control pathways. Immortalization in vitro using virus-derived oncogenes such as large T antigen, E6 and E7, or E1A is based on initial inactivation of the p53 and/or Rb pathways, followed by acquisition of a true immortal phenotype through a telomere-associated genome crisis. hTERT-mediated immortalization, however, escapes the genome instability step but retains the intact checkpoints that limit infinite proliferation (8 , 24 , 58 , 59) . Accumulation of p16INK4A is a hallmark of such a growth restraining mechanism in many cell types, preventing true immortalization by hTERT, and is suggested to be a result of inadequate growth conditions (37 , 38 , 59, 60, 61) . In our study, p16INK4A induction in WI-38/hTERT cells was observed shortly after the cells bypassed the replicative senescence barrier. This induction occurred at the transcriptional level and resulted in protein accumulation. The activation of the p16/pRb pathway could explain the dramatic decrease in the colony formation that we observed when WI-38/hTERTslow cells were seeded at clonal density. Furthermore, ∼5% of the WI-38/hTERTslow cells in mass culture exhibited SA-β-GAL staining, as compared with <1% observed in young WI-38. Increased levels of p16INK4A were shown to induce a senescence-like growth arrest (19 , 36) . On the other hand, variations in levels of p16INK4A expression among individual cells or its inability to completely inhibit CDK4 and CDK6 activity by itself could provide a plausible explanation for the absence of growth arrest in mass culture. A similar p16INK4A induction was observed in several other hTERT-immortalized fibroblast strains (29 , 62) .
p16INK4A inactivation was repeatedly observed in mammary epithelial cells and keratinocytes immortalized by hTERT. Furthermore, its inactivation during hTERT immortalization of two strains of primary human fibroblasts was reported recently (63) . Although INK4A locus silencing could represent a rare event in the hTERT-induced immortalization of human fibroblasts, the apparent growth advantages conferred on cells after its inactivation result in a positive selection process, with rapid outgrowth of INK4A-deficient clones. It is plausible that INK4A locus silencing resulted in a significant acceleration of cell growth, a complete rescue, and even an increase of CFE, as well as loss of contact inhibition. Although we could not distinguish between the individual contribution of either p16INK4A or p14ARF to the observed changes in the growth parameters, both genes act as key regulators of cell growth. Therefore, cells deficient in the expression of either p16INK4A or p14ARF have increased susceptibility to additional transformation (64, 65, 66, 67) .
The strong up-regulation of two oncogenes such as c-myc and Bmi-1, which we observed in close correlation with the INK4A locus silencing, provide additional evidence regarding the premalignant nature of WI-38/hTERTfast cells. Elevated expression of c-myc, for instance, could by itself contribute to the accelerated growth rate we observed. The role of Bmi-1 oncogene in the regulation of INK4A locus expression was suggested by several studies, although the mechanism of its action is still unknown (42 , 44) . The increase in Bmi-1 expression coincides with the decrease in INK4A locus expression, suggesting that in our model, Bmi-1 may serve as a transcriptional regulator of either p16INK4A, p14ARF, or both. Intriguingly, strong cooperation between c-myc and Bmi-1 during lymphomagenesis was demonstrated using an in vivo model (43) . In addition to the possible role of Bmi-1 as a transcription regulator, it seems that in our case, promoter hypermethylation is also involved in p16INK4A silencing. Indeed, we observed that the treatment of WI-38/hTERTfast with a DNA demethylating drug 5Aza-dC partially restored p16INK4A expression. The particular contribution of the Bmi-1-mediated repression and of DNA methylation to INK4A locus silencing in our cellular model is currently under investigation.
It should be noted that a number of genetic and epigenetic events such as activation of the c-myc oncogene and the inactivation of INK4A locus were recently reported to be associated with hTERT-induced immortalization of human mammary and adenoid-derived epithelial cells (39 , 68) . Furthermore, hTERT-associated growth acceleration and profound changes in the transcription pattern were noted (69 , 70) . Studying mice that constitutively express high levels of telomerase in basal keratinocytes suggested an important role of telomerase in promoting tumorigenesis in vivo. Those mice exhibit a greater incidence of spontaneous and carcinogen-induced tumors than wild-type controls (71 , 72) . In general, our findings are in agreement with those reported for MRC-5 lung fibroblasts immortalized by telomerase expression (73, 74, 75) . However, in those studies, no extensive molecular characterization of the pRb and p53 pathways was presented. Therefore, it is impossible to apply our conclusions to other types of human fibroblasts.
Loss of intact p53- and pRb-mediated cell cycle checkpoints are common events during malignant transformation. According to our findings and in agreement with previously published data (28 , 29) , ectopic expression of hTERT does not affect the p53-mediated DNA damage cell cycle checkpoint or the pRb-mediated quiescence response. Our cytogenetic analysis confirms that the p53-mediated cell cycle checkpoint is intact and suggests that hTERT overexpression and the maintenance of telomeres do not lead to accumulation of genomic aberrations characteristic of cancer cells.
Compelling evidence for the premalignant nature of INK4A-deficient hTERT-immortalized cells is provided by their response to an oncogenic mutant of Ras. Induction of irreversible growth arrest with features of replicative senescence was repeatedly reported for primary cells with intact p53 and pRb pathways in response to mutant Ras overexpression (51, 52, 53 , 56) . This arrest prevents Ras-induced transformation and serves as a barrier against oncogene-driven tumorigenesis. In stark contrast to the irreversible growth arrest observed in primary and in WI-38/hTERTslow cells after H-Ras overexpression, the INK4A-deficient WI-38/hTERTfast cells expressing Ras resumed proliferation and became transformed, as judged by their ability to grow in an anchorage-independent manner. Although there are different requirements for ARF versus p16INK4A expression in mouse and human cells controlling their sensitivity to Ras-mediated transformation (56 , 57 , 62 , 76, 77, 78, 79) , the complete silencing of both genes seems to underlie the transformation by Ras as observed in WI-38/hTERTfast cells. Spectral karyotype analysis of WI-38/hTERT/H-Ras/GSE56 cells revealed a strong destabilizing effect of those oncogenes on chromosomal integrity, in addition to their known effect on cell growth parameters.
Our data, taken together with previous studies regarding the genome integrity during the different stages of immortalization, suggest that hTERT-immortalized human fibroblasts maintain a stable diploid karyotype for an extended time period. Stable changes in the expression of genes such as p16INK4A and p14ARF could be attributed to epigenetic events. However, the unlimited proliferative potential conferred to cells by hTERT, in concert with defects in INK4A locus-dependent failsafe programs, provide a suitable background for rapid accumulation of additional genetic aberrations as illustrated by the oncogene overexpression demonstrated in our study.
In summary, our studies of hTERT-induced immortalization of WI-38 lung embryonic fibroblasts provide strong evidence for the positive selection of potentially malignant genetic alterations during prolonged culture in vitro. Although hTERT by itself does not induce cell transformation, our results emphasize the need for careful consideration of safety before hTERT-immortalized cells could be used for cell therapy.
We thank Dr. Irit Bar-Am (Applied Spectral Imaging Ltd.) for assistance with spectral karyotype, Dr. Jerry Shay (The University of Texas Southwestern Medical Center) for the hTERT plasmid, Dr. Andrei Gudkov (Lerner Research Institute, Cleveland Clinic Foundation) for the GSE56 plasmid, and Dr. Moshe Oren (Weizmann Institute) for fruitful discussions.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This study was supported in part by a grant from the Israel-USA Binational Science Foundation (to C. C. H., V. R.). V. R. holds the Norman and Helen Asher Professorial Chair in Cancer Research at the Weizmann Institute.
↵2 Supplemental data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org).
↵3 To whom requests for reprints should be addressed, at Department of Molecular Cell Biology, Weizmann Institute of Science, Rehovot 76100, Israel. Phone: 972-8-9344501; Fax: 972-8-9465265; E-mail:
↵4 The abbreviations used are: HDF, human primary diploid fibroblast; SA-β-GAL, senescence-associated β-galactosidase; CDK, cyclin-dependent kinase; PDL, population doubling; TRF, terminal restriction fragment; RT-PCR, reverse transcription-PCR; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; BrdUrd, bromodeoxyuridine; TRFL, telomere restriction fragment length; CFE, colony-forming efficiency; 5-AzaC, 5-aza-2′-deoxycytidine.
- Received May 20, 2003.
- Revision received August 13, 2003.
- Accepted August 19, 2003.
- ©2003 American Association for Cancer Research.