Chromosomal instability is correlated with telomere erosion ... - Nature

Oncogene (1998) 16, 1825 ± 1838  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc

Chromosomal instability is correlated with telomere erosion and inactivation of G2 checkpoint function in human ®broblasts expressing human papillomavirus type 16 E6 oncoprotein Leonid Filatov, Vita Golubovskaya, John C Hurt, Laura L Byrd, Jonathan M Phillips and William K Kaufmann Department of Pathology and Laboratory Medicine, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-7295, USA

Cell cycle checkpoints and tumor suppressor gene functions appear to be required for the maintenance of a stable genome in proliferating cells. In this study chromosomal destabilization was monitored in relation to telomere structure, lifespan control and G2 checkpoint function. Replicative senescence was inactivated in secondary cultures of human skin ®broblasts by expressing the human papillomavirus type 16 (HPV-16) E6 oncoprotein to inactivate p53. Chromosome aberrations were enumerated during in vitro aging of isogenic control (F5neo) and HPV16E6-expressing (F5E6) ®broblasts. We found that structural and numerical aberrations in chromosomes were signi®cantly increased in F5E6 cells during aging in vitro and ¯uorescence in situ hybridization (FISH) analysis using chromosome-speci®c probes demonstrated the occurrence of rearrangements involving chromosome 4 and 6 in genetically unstable F5E6 cells. Flow cytometry and karyotypic analyses revealed increased polyploidy and aneuploidy in F5E6 cells only at passages 416, although these cells displayed defective mitotic spindle checkpoint function associated with inactivation of p53 at passages 5 and 16. G2 checkpoint function was con®rmed to be gradually but progressively inactivated during in vitro aging of E6-expressing cells. Aging of F5neo ®broblasts was documented during in vitro passaging by induction of a senescence-associated marker, pH 6.0 lysosomal b-galactosidase. F5E6 cells displayed extension of in vitro lifespan and did not induce b-galactosidase at high passage. Erosion of telomeres during in vitro aging of telomerase-negative F5neo cells was demonstrated by Southern hybridization and by quantitative FISH analysis on an individual cell level. Telomeric signals diminished continuously as F5neo cells aged in vitro being reduced by 80% near the time of replicative senescence. Telomeric signals detected by FISH also decreased continuously during aging of telomerasenegative F5E6 cells, but telomeres appeared to be stabilized at passage 34 when telomerase was expressed. Chromosomal instability in E6-expressing cells was correlated (P50.05) with both loss of telomeric signals and inactivation of G2 checkpoint function. The results suggest that chromosomal stability depends upon a complex interaction among the systems of telomere length maintenance and cell cycle checkpoints. Keywords: chromosome; instability; telomere; cell cycle; checkpoint

Correspondence: WK Kaufmann Received 14 August 1997; revised 4 November 1997; accepted 5 November 1997

Introduction Chromosomal instability is common in cancer cells (Holliday, 1989; Hartwell and Kastan, 1994). Not only do cancers display abnormalities of chromosomes, be they polyploidy, aneuploidy, interstitial deletions and ampli®cations, or a single marker chromosome, but malignant cells also appear to acquire such abnormalities at increased rates in comparison to their normal progenitors. The mechanisms of genetic instability in cancer cells are, therefore, of considerable interest. Chromosome instability in human cells may be caused by defects in various elements of DNA metabolism including replication, chromosome segregration, repair and recombination processes (Hartwell, 1992; Cohen and Levy 1989; Coquelle et al., 1997). Chromosomal aberrations and alterations in DNA ploidy can be induced by treatments with various drugs that damage DNA, including chemotherapeutic agents and chemical carcinogens, or by exposure to ionizing and ultraviolet radiations. Remarkably, expression of viral oncoproteins that inactivate tumor suppressor gene functions also induces chromosomal aberrations (Stewart and Bacchetti, 1991; Chang et al., 1997), implying that chromosome stability is preserved by tumor suppressor gene expression. The tumor suppressor genes p53 and pRB serve within checkpoint circuits that regulate cell division. Cell cycle checkpoints represent positions of control that ensure the completion of dependent events in the cell division cycle and provide more time for DNA repair before DNA replication and mitosis (Hartwell and Kastan, 1994). Two DNA damage-responsive checkpoints act to delay the G1?S and G2?M cycle phase transitions (Kaufmann and Paules, 1996). Inactivation of p53 ablates G1 checkpoint function and is associated with gene ampli®cation and chromosomal instability in human ®broblasts (Yin et al., 1992; Livingstone et al., 1992; White et al., 1994). Ataxia telangiectasia cells that are defective in both G1 and G2 checkpoint functions display enhanced frequencies of spontaneous and radiation-induced chromosomal aberrations (Taylor et al., 1976; Zampetti-Bosseler and Scott, 1981; Ejima and Sasaki, 1986). Defects in G2 checkpoint function were associated with enhanced frequencies of radiationinduced chromosome breaks in a panel of human cancer lines (Schwartz et al., 1996). Cell cycle checkpoints, consequently, appear to preserve genetic stability and suppress carcinogenesis (Hartwell, 1992; Kaufmann and Kaufman, 1993).

Chromosomal instability in HPV-16E6-expressing fibroblasts L Filatov et al

1826

The clastogenic eects of oncogenic viruses have been related in part to their abilities to bind to and inactivate tumor-suppressor genes (Stewart and Bacchetti, 1991; Chang et al., 1997). Certain viruses, such as SV40, adenovirus and oncogenic strains of human papillomavirus (HPV), alter the functions of the protein products of p53 and pRB. In the case of HPV-16, the E6 gene product targets p53 for ubiquitinmediated proteolysis (Schener et al., 1990; Galloway et al., 1994). A strain of HPV with low oncogenicity (HPV-6) expresses an E6 protein that is unable to target p53 for proteolysis and fails to inactivate the G1 checkpoint (Galloway et al., 1994). Carcinogenesis by HPV-16 therefore appears to depend upon inactivation of p53 function by its E6 gene. HPV-16E6-immortalized human urothelial cells displayed high frequencies of unstable chromosomal aberrations and stable marker chromosomes as an example of the chromosomal instability that develops in HPV-16E6-expressing cells (Rezniko et al, 1994). By using an amphotropic retrovirus to induce synchronous expression of HPV16E6 in neonatal human skin ®broblasts, it was possible to monitor the kinetics of cell cycle checkpoint inactivation and chromosomal destabilization under the in¯uence of a viral oncoprotein. Expression of the HPV-16E6 oncoprotein in neonatal human diploid ®broblasts inactivated the G1 checkpoint (Dulic et al., 1994), but the G2 checkpoint was unaected initially (Paules et al., 1995; Levedakou et al., 1995) and cells were normal cytogenetically (White et al., 1994). However, as E6-expressing cells aged in vitro they displayed increased frequencies of chromosomal abnormalities including telomere associations, chromosomal aberrations and aneuploidy (White et al., 1994; Kaufmann et al., 1997). This chromosomal instability was associated with concurrent attenuation of G2 checkpoint function (Paules et al., 1995; Kaufmann et al., 1997). The results suggested that oncogene-mediated inactivation of p53 can lead to subsequent genetic alterations during cellular aging that deregulate the G2 checkpoint and induce chromosomal instability. Chromosomal instability also has been associated with telomere shortening during cellular aging in vitro. Progressive erosion of telomeres in precrisis SV40 virus-transformed cells and p53-de®cient Li ± Fraumeni Syndrome (LFS) ®broblasts was associated with increased numbers of telomere associations, dicentric chromosomes, and ring chromosomes (Counter et al., 1992; Rogan et al., 1995). Telomeres are T2AG3 repeats complexed with specialized proteins that are located at the ends of all eukaryotic chromosomes. Telomeres are essential for genetic stability. Telomeres dierentiate chromosome ends from double-stranded breaks and protect them from aberrant recombination and nuclease degradation (McClintock, 1941; for reviews see Healy, 1995; Ishikawa, 1997). In normal human cells that lack expression of the enzyme telomerase, the ends of chromosomes gradually shorten with advancing age in vivo and with increasing number of passages in vitro due to the inability of DNA polymerases to replicate the 3' ends of linear DNA (Olovnikov, 1971; Healy, 1995). Telomere shortening below a certain length was proposed to induce irreversible cell cycle arrest known as replicative cell senescence (Harley et al., 1990). Chromosome ends with critically eroded

telomeres may be sensed as irreparable DNA damage signals by the p53-dependent G1 checkpoint (Dulic et al., 1994; Kaufmann and Paules, 1996). Telomere length can be maintained in immortal cells, stem cells and germ-line cells by a ribonucleoprotein enzyme, telomerase, which adds new telomeric repeats to chromosome ends (for review see Greider and Blackburn, 1985; Kim et al., 1994). An alternative mechanism for telomere maintenance in immortal human cell lines may involve recombination (Bryan et al., 1995; Rogan et al., 1995). Recently a human telomeric-repeat binding factor TRF1 was shown to be involved in telomere length regulation (van Steensel and de Lange, 1997). It was proposed that the binding of TRF1 controls telomere length in cis by inhibiting the action of telomerase at the ends of individual telomeres (van Steensel and de Lange, 1997). To explore further the mechanisms of chromosomal instability in HPV-16E6-expressing cells that lack the replicative senescence checkpoint function, we have monitored cellular aging during in vitro proliferation by assay of senescence-associated b-galactosidase expression and by quantitative analysis of telomere structure. In an attempt to identify the origin of dierent chromosome aberrations we analysed chromosomal instability in relation to telomere structure and G2 checkpoint function. The development of chromosomal abnormalities was correlated (P50.05) with both telomere erosion and attenuation of G2 checkpoint function in HPV-16E6-expressing cells. Chromosomal stability appears to depend upon a complex interaction among the systems of telomere length maintenance and cell cycle checkpoints.

Results Age-dependent appearance of chromosomal instability in F5E6 ®broblasts We used two human ®broblast lines, F5neo and F5E6, to study chromosomal destabilization during in vitro aging. F5E6 cells lacked G1 checkpoint function due to inactivation of p53 by the HPV-16E6 oncoprotein (Dulic et al., 1994; Kaufmann et al., 1997). Degradation of p53 protein was con®rmed by Western blotting with p53-speci®c antibody (results not shown) and inactivation of G1 checkpoint function was con®rmed by ¯ow cytometry after ionizing radiation treatment (Kaufmann et al., 1997). Normal F5neo and p53negative F5E6 ®broblasts displayed in vitro population expansion typical of cells of these types (Shay et al., 1993; White et al., 1994). F5neo ®broblasts underwent 65 population doublings between passages 3 and 27, after which senescence occurred. F5E6 cells had undergone 66 population doublings by passage 22 and at passage 27 (population doubling level 88) F5E6 cells were at the stage of crisis in which population expansion was stable even though cells were dividing. Unexpectedly, F5E6 cells resumed population expansion by passage 34 and threafter population expansion was continuous. In a previous analysis of these cells (Kaufmann et al., 1997) G2 checkpoint function and chromosomal aberrations were charted continuously as F5neo and F5E6 cells aged through their in vitro proliferative life


spans. This experimental design required that cytogenetic and checkpoint analyses be performed over a 6 month interval as cells were passaged weekly. During this analysis cells were cryopreserved at various passage levels. Subsequent re-establishment of low and intermediate passage preparations enabled synchronous assessment of chromosomal aberrations and G2 checkpoint function in cells of various ages. Additional evaluation of a senescence marker and telomere structure permitted evaluation of the association between cellular aging and chromosomal instability. We ®rst quanti®ed chromosomal aberrations in the cell lines at various phases of their in vitro proliferative life span to determine the reproducibility of our previous measurement (Kaufmann et al., 1997). F5E6 cells, but not F5neo cells, displayed signi®cant age-dependent abnormalities of chromosome number and structure. The results of analyses of chromosome numbers are summarized in Figure 1 and Figure 2. F5neo cells had diploid numbers of chromosomes (46+1) at all passages (Figure 1a), as did F5E6 cells at passages 5 and 16 (Figure 1b). At passages above 16 F5E6 cells with near 2n chromosome number displayed

a

increased numbers of aneuploid metaphases with both losses and gains of chromosomes (Figure 1b). The F5E6 line also had increased percentages of cells with 3-4n chromosome numbers (hypotetraploids and tetraploids) at passages 22, 27 and 34 (Figure 2b). The passage 27 F5E6 cells in crisis displayed severely aberrant ploidies with increased percentages of cells with 5-13n chromosome numbers (Figure 2b and Figure 3b). The reduced fraction of highly polyploid metaphases in the postcrisis F5E6 cells at passage 34 probably re¯ects the emergence of one or a few clonal sublines that maintain hypodiploid and hypotetraploid karyotypes. F5neo cells maintained 2n chromosome content up to near senescence at passage 27. Structural chromosomal abnormalities were infrequent in F5neo cells at all passages. Among 50 diploid metaphases sampled at each passage of F5neo cells, the percentage with a chromosome aberration was 2 ± 6% (results not shown). F5E6 ®broblasts exhibited similar low frequencies of aberrations in diploid metaphases at low passage levels (5 and 16) (Table 1). Starting with passage 22 the numbers of chromosomal aberrations in near-diploid F5E6 cells were progressively and signi®cantly increased. Examples of the types of aberrations seen in F5E6 are shown in Figure 3. Structural aberrations included telomere associations, dicentrics, breaks, exchanges, ring chromosomes and

a

b

b

Figure 1 Distribution of F5neo and F5E6 cells with near diploid chromosome numbers. (a) F5neo cells at passages 5, 16, 22 and 27 (cummulative population doubling levels 8, 42, 50 and 65, respectively). (b) F5E6 cells at passages 5, 16, 22, 27 and 34 (population doubling levels 8, 47, 66, 88 and 94 respectively). 50 metaphases were analysed for each passage number, except for passage 34, where 33 metaphases were analysed. For chromosome number 46: P50.001 in F5E6p22, 27 and p34 vs F5E5p5 cells. For chromosome numbers, 37 ± 47: P50.03 in F5E6p22, 27 and 34 vs F5E6p5 cells. For chromosome numbers, 47 ± 52: P50.002 in F5E6p27 vs F5E6p 5 cells (Chi-square test)

Figure 2 Polyploidy analysis in F5neo and F5E6 cells. (a) F5neo cells at passages 5, 16, 22 and 27. (b) F5E6 cells at passages 5, 16, 22, 27 and 34. 500 metaphases were examined for analysis of polyploidy for each passage level except for p34, where 50 metaphases were examined. For 3 ± 4 N, P50.0001 in F5E6p22, 27 and 34 vs F5E6p5. For 5 ± 13 N, P50.0001 in F5E6 p27 vs F5E6 p5 (Chi-square test)

1827


1828

a

b

c

d

Figure 3 Structural and numerical chromosome aberrations in F5E6 ®broblasts at passage 27. (a) Part of a Giemsa-stained metaphase with a chromatid break (single arrow), chromatid exchange (double arrow) and telomere association (triple arrow). (b) Polyploid DAPI-stained metaphase with dicentrics (arrows), ring chromosome (arrowhead) and chromosome breaks (double fragments). (c,d) FISH with directly labeled chromosome-speci®c probes (chromosome 4 ± spectrum green, chromosome 6 ± spectrum orange). (c) Metaphase showing translocation involving part of chromosome 4 (small arrowhead), normal chromosome 4 (arrow) and deleted isodicentric chromosome 4: idic(4)(q32?):4pter?cen?4q32?::4q32??cen?4pter (large arrowhead), and a translocation involving part of chromosome 6 (wide arrow). (d) Tetraploid cell with four normal chromosomes 6 (red), one normal chromosome 4 (green, arrow) and four translocations involving chromosome 4 (arrowheads)


1829

Table 1 Chromosomal aberrations in F5E6 ®broblasts Passage p5 p16 p22 p27 p34

Telomere associations Dicentrics 0 1 2 4** 0

0 1 3 5** 0

Aberrations Breaks Chm Chd 0 1 1 4 3

1 1 1 7 3

Chd Exch

Ring Chm

0 0 0 2 0

0 0 1 2 0

Total (freq.) 1 4 8 24 6

(0.02) (0.08) (0.16) (0.48) (0.18)

% aberrant cells 2 8 14*a 28*b 15*c

Abbreviations: Chm ± chromosome, Chd ± chromatid, P ± passage, Exch ± exchange, Freq. ± frequency. Fifty metaphases with diploid chromosome numbers (46+1) were examined for every passage level (except, for passage 34, where 33 metaphases were analysed). *For total % aberrant cells: P50.05 vs F5E6p5 (Chi-square test). aat p22, 1 cell (2%) had 2 aberrations, dicentric+Chd break. bAt p27, 4 cells (8%) had 2 aberrations: 2 cells with Ring Chm+Chd break, 1 cell with Chd exchange+Chm break, 1 cell with telomere association+Chd break; 3 cells (6%) had 3 aberrations: 1 cell with 2 dicentrics+Chm break, 1 cell with telomere association+dicentric+chd break, 1 cell with 2 chd breaks+dicentric. cAt p34, 1 cell (2%) had Chm break+Chd break. **For telomeric associations and dicentrics: P50.01 vs F5E6 p5 and p34 (Chi-square test)

translocations. The numbers of telomere associations and dicentrics increased progressively with aging of F5E6 ®broblasts (Table 1). No such aberrations were found at passage 5 whereas at passage 27 ®ve dicentrics and four telomere associations were observed. F5E6 cells also displayed a similar age-dependent increase in chromatid chromosome breaks, ring chromosomes, and exchanges between passages 5 and 27. The postcrisis F5E6 at passage 34 displayed increased frequencies of chromosome and chromatid breaks, but no telomeric associations or dicentrics were seen in the 33 diploid metaphases that were scored. The percentages of F5E6 cells with telomeric associations or dicentrics were signi®cantly dierent at passages 27, in which 7 of 50 of metaphases were positive, and passage 34, in which no metaphases were positive (Chisquare, P50.05). Rearrangements of chromosomes 4 and 6 in F5E6 ®broblasts at passage 27 G-banding of F5E6 cells was used for preliminary detection of rearranged chromosomes (not shown). Observation of rearrangements involving chromosomes 4 and 6 led us to perform FISH with chromosomespeci®c probes. Two near-tetraploid metaphases with examples of rearrangements of these chromosomes are shown in Figure 3c and d. The rearrangements with chromosome 4 included one deleted isodicentric (Figure 3c) and several dierent translocations (Figure 3d). A rearrangement involving a portion of chromosome 6 is shown in Figure 3c. Mitotic spindle checkpoint function The analyses of chromosome numbers (Figure 2) revealed increased frequencies of polyploid metaphases in aging F5E6 cells. Polyploidization can imply inactivation or deregulation of mitotic spindle checkpoint function (Cross et al., 1995; Minn et al., 1996; DiLeonardo et al., 1997). The mitotic spindle checkpoint was assayed by incubating cells in colcemid to inhibit microtubule polymerization and quantifying initiation of DNA synthesis after 24 h (Cross et al., 1995; DiLeonardo et al., 1997) (Figure 4). In the absence of colcemid, F5neo and F5E6 cells at passage 9 displayed quite similar pro®les of incorporation of BrdU, with 1 ± 2% of cells in the tetraploid S, G2 and

M compartment. When treated with colcemid, F5E6 cells at passage 9 showed a sevenfold accumulation of cells in the diploid G2/M compartment (from 7 ± 51%) and increased polyploidization (22% of cells in tetraploid S, G2 and M compartment). F5neo controls also displayed sevenfold accumulation in the diploid G2/M compartment during incubation with colcemid but only 2% of cells were in the tetraploid S, G2 and M compartment (Figure 4). Passage 25 F5E6 cells displayed 16% tetraploid cells when incubated with colcemid (data not shown). However, in the absence of colcemid treatment the high passage F5E6 cells also displayed increased fractions of cycling cells with tetraploid DNA content (14% of cells in tetraploid S, G2 and M) as compared to F5neo controls (Figure 4). The 14% of passage 25 F5E6 cells in the tetraploid S, G2 and M compartment may be matched by an equal or larger percentage of tetraploid G1 phase cells that are superimposed on the diploid G2/M compartment. Thus, the ¯ow cytometry analyses supported the cytogenetic data showing increasing numbers of polyploid F5E6 cells during in vitro aging. Inactivation of G2 checkpoint function in F5E6 ®broblasts during aging As a further control G2 checkpoint function was quanti®ed by enumerating the fractions of cells in mitosis 2 h after treatment with g-irradiation (Table 2). F5neo cells at passage level 5 had a strong G2 checkpoint response with no cells escaping radiationinduced G2 delay. G2 checkpoint function remained unchanged as F5neo ®broblasts aged from passage 5 to 27. In F5E6 ®broblasts at passage 5 G2 checkpoint function was normal, as only 4% of G2 phase cells entered mitosis 2 h after 1.5 Gy of g-rays. As F5E6 cells aged in vitro they displayed progressive attenuation of G2 checkpoint function in response to g-irradiation, as was shown previously (Paules et al., 1995; Kaufmann et al., 1997). When F5E6 cells were in crisis at passage level 27, 100% of irradiated cells excaped the G2 checkpoint and entered mitosis 2 h after 1.5 Gy. Aging of F5neo ®broblasts con®rmed by b-galactosidase assay To monitor the kinetics of cellular aging during in vitro passaging of the ®broblast lines, we assayed the


1830

senescence-associated marker, pH 6.0 b-galactosidase (Dimri et al., 1995). When tested at passage 5, both lines displayed few b-galactosidase-positive cells (Figure 5). As cells aged they displayed dierent levels of expression of b-galactosidase. Virtually all F5neo cells expressed pH 6.0 b-galactosidase activity at passage 27. In contrast, few if any F5E6 cells expressed this marker of senescence when in crisis at passage 27 (Figure 5). Expression of telomerase in F5E6 ®broblasts only at late passages We assayed for telomerase activity in F5neo ®broblasts at passages 10 and 26 and F5E6 ®broblasts at passages 10, 26 and 34 using TRAP (Figure 6). Telomerase activity was not detected at passage 10 in F5neo and F5E6 cells, while at passage 26 some telomerase activity was detected in F5E6 but not in F5neo cells. Telomerase activity was relatively high at passage 34 in F5E6 cells.

Terminal restriction fragment length analysis in F5neo and F5E6 ®broblasts Southern hybridization analysis of the telomereassociated terminal restriction fragment (TRF) is shown in Figure 7. TRF length and hybridization intensity decreased with cell aging in vitro. At passage 9, TRF length in F5neo and F5E6 cells did not dier substantially. As F5neo cells aged in culture, the TRF shortened modestly and the intensity of hybridization with the telomere-speci®c probe was diminished. The TRF length decreased more extensively in F5E6 cells than in F5neo cells (Figure 7). At passage 34, the greatest mass of telomere-probe-hybridizing DNA in F5E6 cells was less than 2 kb in length (Figure 7). The TRF length remained constant at this level in F5E6 cells, as it did not decrease further at higher passages (data not shown).

Figure 4 Attenuation of mitotic spindle checkpoint function in F5E6 ®broblasts. F5neo and F5E6 cells at passages 9 and 25 in logarithmic growth phase were incubated for 24 h with and without 0.27 mM colcemid. Addition of BrdU for the last 2 h of colcemid incubation allowed the determination of the fraction of cells that were synthesizing DNA. Fixed cells were processed for two-parameter ¯ow-cytometric analysis of DNA content (propidium iodide) vs incorporation of BrdU (FITC). The small boxes enclose diploid cells in G2/M and tetraploid cells in G0/G1, the large boxes enclose tetraploid cells in S, G2 and M. The percentages of cells in these boxes are indicated

Table 2 Attenuation of G2 checkpoint function in F5E6 cells during in vitro aging Passage 5 16 22 27

F5neo sham % g-Irradiated 1.8 1.6 1.1 0.15

0 0.05 0 0

Mitotic indexa % escaping G2 F5E6 checkpointb sham % g-Irradiated 0 3 0 0

3.4 3.4 10.4 5.5

0.15 0.8 4.8 5.8

% escaping G2 checkpointb 4 24* 46* 100*

a The percentage of cells in mitosis 2 h after sham treatment or irradiation with 1.5 Gy g-rays (number mitotic cells/ total number cells counted, n=2000). bThe percentage represents the fraction of irradiated G2 phase cells that failed to delay entry into mitosis 2 h after irradiation with 1.5 Gy g-rays. *P50.01 vs F5E6 p5


1831

Figure 5 Expression of b-galactosidase in aged F5neo but not F5E6 cells. Assay of pH 6.0 b-galactosidase was done as described in Materials and methods. (a) F5neo cells at passage 5, (b) F5E6 cells at passage 5. (c) F5neo cells at passage 27, (d) F5E6 cells at passage 27. Phase-contrast photomicrographs were image-captured and a composite generated using Macintosh Photoshop software

In situ analysis of telomere erosion in F5neo and F5E6 cells during in vitro aging In order to establish the kinetics of telomere shortening during in vitro aging on an individual cell basis, we used FISH with a human telomere probe, which is a direct method to determine telomere length (Henderson et al, 1996; Therkelsen et al., 1995; Lansdorp et al., 1996). The number and size of telomeric signals as well as their intensity are functions of telomere length (Lansdorp et al., 1996). The total number of visible telomeric signals over diploid metaphases (FITCpositive dots at chromatid termini) decreased with in vitro aging at similar rates in normal and E6-expressing ®broblasts (Figure 8a ± g, Table 3). However, starting with passage 22 the dierence in the number of telomeric signals between F5neo and F5E6 ®broblasts was signi®cant (P50.05) (Figure 9). At passage 27 near the time of replicative senescence, the number of telomeric signals detected in mitotic F5neo cells was reduced by 80% relative to the signals observed at passage 5. The average number of FITC-positive signals in F5E6 cells was reduced still further from 159 (at passage 5) to 21 (at passage 27), representing a decrease of 87% when F5E6 cells were in crisis (Figure 9). A method for enumeration of telomere intensity based on photodetection yielded an equivalent result (R2=0.995, P50.01) (Table 3). Correlations between inactivation of G2 checkpoint function, telomere erosion and chromosomal instability in E6-expressing ®broblasts The following data were analysed for correlations: percent of cells evading the G2 checkpoint (Table 2);

percent of diploid metaphases with structural chromosomal aberrations (Table 1); percent of polyploid (42n) metaphases (Figure 2b); and mean number of telomere-speci®c spots in diploid metaphases (Figure 9, Table 3). All but one comparisons indicated signi®cant correlations (P50.05). The only pairwise comparison that was not correlated with a high probability was the percent of polyploid metaphases versus telomere spot number. The greatest correlation was between the percentages of cells escaping the G2 checkpoint and diploid cells with chromosomal aberrations (P=0.0005). Discussion Chromosomal instability in HPV16E6-expressing human skin ®broblasts lacking p53-dependent G1 checkpoint function was correlated with both telomere erosion and inactivation of G2 checkpoint function. By quantitative FISH analysis on an individual cell basis it appeared that telomeres eroded at similar rates in diploid F5neo and F5E6 cells, with approximately 80% loss of telomeric sequences at the time of replicative senescence in F5neo cells. F5E6 cells bypassed the replicative senescence checkpoint, con®rmed by expression of 20 additional population doublings before crisis (Shay et al., 1993; Rogan et al., 1995), lack of expression of the senescence-associated marker, pH 6.0 lysosomal b-galactosidase (Dimri et al., 1995) and more severe loss of telomeric sequences. G2 checkpoint function was normal soon after expression of HPV-16E6 but progressively degraded during aging of F5E6 cells. F5neo cells had no changes of karyotype during in vitro aging and no


F5E6

p34 F5E6

F5

p27 F5E6

F5

p21 F5E6

p9 F5

λ/Hind lll

Lysis Buffer

F5E6

p34 F5E6

p26

F5

F5E6

F5

p10

MDAH041

1832

23.1 —

9.4 — 6.5 —

4.4—

2.3 — 2.0 —

Figure 6 Telomerase activity was examined in F5neo and F5E6 cells. Telomerase activity was examined in F5neo cells at passage 10 and 26 and in F5E6 cells at passages 10, 26 and 34 by PCRbased TRAP assay as described in Materials and methods. MDAH 041 (an immortal LFS ®broblast line) was used as a positive control and lysis buer was used as a negative control

loss of G2 checkpoint function. In contrast, F5E6 cells at low passage (5) had normal karyotype, but starting with passage 22, the number of cells with structural and numerical chromosome aberrations increased progressively. The p53-dependent mitotic spindle checkpoint was defective in F5E6 cells starting in early passages when chromosomes were normal, indicating that this checkpoint alteration also was insucient for chromosome instability in F5E6 cells. Thus, signi®cant correlations were observed among telomere erosion, inactivation of G2 checkpoint function, and structural and numerical chromosome abnormalities during the aging of cells lacking G1 checkpoint function. The results that the HPV-16E6-expressing ®broblasts displayed extended lifespan in vitro at late passages are consistent with data on spontaneous immortalization of LFS ®broblasts (Rogan et al., 1995). It has been shown that loss of wt p53 by either mutation or deletion results in a ®nite increase in the in vitro proliferative potential of human ®broblasts (Shay et

Figure 7 Terminal restriction fragment length analysis in F5neo and F5E6 cells. Genomic DNA was isolated from F5 and F5E6 cells at the passages indicated and digested with HinfI and RsaI enzymes. The digested DNA was separated by electrophoresis on a 0.8% agarose gel and hybridized to digoxigenin-labeled (TTAGGG)3 probe. Molecular weight markers (Lambda DNA/ HindIII digests) are indicated on the left. The image is a composite of two dierent exposures to optimize visualization of hybridization in the high passage cells

al., 1993; Bond et al., 1994; White et al., 1994). A previous analysis by Shay et al. (1993) suggested that inactivation of both p53 and pRB was required for immortalization of human ®broblasts. However, wild type pRB was observed in one line of spontaneously immortalized LFS ®broblasts (Rogan et al., 1995). These immortal ®broblasts had no detectable p16INK4 protein (possibly equivalent to inactivation of pRB) and expressed chromosome destabilization of the types described here (Rogan et al., 1995). Other genetic events leading to elevated expression of cyclins, cyclindependent kinases (CDK) and CDK inhibitors may override pRB and contribute to escape from senescence (Grana and Reddy, 1995). The sequence of alterations in chromosome structure and number observed in F5E6 cells closely resembled those previously documented during stages of immortalization of human ®broblasts induced by SV40 virus (Counter et al., 1992) and epithelial cells induced by HPV-16E6 (Klingelhutz et al., 1994). In both of these examples, severe chromosomal instability developed during cellular


aging, peaking at the time of crisis, when most cells died. Post-crisis lines expressing telomerase and stable telomeres also appeared to have partially stabilized chromosomes.

Telomeres eroded with age in vitro in both F5neo and F5E6 cells, reaching a stable stage at passage 34 in F5E6 cells, when telomerase was expressed. Klingelhutz et al. (1994) measured telomere length in relation

Figure 8 FISH analysis of telomere erosion in F5neo and F5E6 cells. Digoxigenin-labeled `all-human-telomeric' probe was used for FISH analysis. FISH analysis was performed under identical conditions for cells at each passage level. After detection with antidigoxigenin-FITC, chromosomes were counterstained with propidium iodide. Fluorescence photomicrographs were imagecaptured and a composite generated using Macintosh Photoshop software. (a,b,c) F5neo metaphases from passages 5, 16 and 27, respectively. (d,e,f,g) F5E6 metaphases from passages 5, 16, 27 and 34, respectively

1833


1834

Figure 9 Telomere erosion dynamics in F5neo and F5E6 cells. Metaphases were scanned after FISH with digoxigenin-labeled telomere probe. The number of telomeric FITC-signals was expressed as a function of cell aging. The values shown are means+standard deviations (n=10). Telomeres eroded significantly in F5E6 cells as compared to F5neo cells at high passages (p22 and 27, Student's t-test, P50.05). FITC-labeled metaphases were also evaluated by a photomicroscopic method, which determined the number of seconds needed to achieve sucient exposure (Table 3)

Table 3 Quantitation of telomeric FITC signals during cellular aging Cell line/passage F5neo p5 F5E6 p5 F5neo p16 F5E6 p16 F5neo p22 F5E6 p22 F5neo p27 F5E6 p27 F5E6 p34

Mean # of signals (+s.d.)

Exposure time (s+s.d.)**

160+7 159+7 109+7 108+4 69+6 61+6* 31+6 22+7* 16+6

91+7 68+5 132+7 136+7 191+9 193+8 230+9 235+6 245+8

*P50.05 vs F5neo at same passage. **Exposure time is inversely proportional to the intensity of FITC-telomeric signals in the metaphase. Linear regression of mean # of signals against exposure time yielded a correlation coecient (R2) of 0.995, P50.001

to immortalization of cervical epithelial cells and found that telomeres gradually shortened in the pre-senescence normal and precrisis E6-expressing cells. In contrast to precrisis cells, E6/E7-immortalized cells generally showed elongated telomeres suggesting that arrest of telomere shortening may be important for HPV-associated immortalization. In our study the TRF length had decreased extensively in F5E6 cells near the time of crisis. The TRF has been shown to be composed of telomere and telomere-like sequences as well as non-telomeric sequences (Counter et al., 1992; Henderson et al., 1996). Thus, TRF size distributions re¯ect chromosomal and cellular heterogeneity in both T2AG3 and non-T2AG3 components. In situ hybridization with a telomere-speci®c probe quanti®ed changes in telomere size on an individual cell level in the diploid population of dividing cells. This method is particularly useful for small populations of cells and when applied to the F5neo and F5E6 ®broblasts, quantitative telomere FISH showed a continuous

decrease in intensity and detectability of telomeric signals during in vitro aging (Table 3). Interchromosomal heterogeneity in telomere signals (Landsorp et al., 1996) in aged cells suggests that shortening of certain chromosome telomeres below a critical length could trigger cell senescence (Healy, 1995). E6-expressing cells did not undergo cell senescence, but as telomeres continued to erode in the highly proliferative aging population, the frequency of telomere associations and dicentric chromosomes increased until cells reached crisis at or about passage level 27. Chromosomal endto-end associations and fusions were found to increase progressively over several generations of mTR (mouse telomerase RNA) knockout mice, as telomeres were eroded (Blasco et al., 1997). Our data also indicate that telomere shortening is associated with chromosome destabilization in human cells lacking p53-dependent G1 checkpoint function. Several types of chromosome aberrations were observed in E6-expressing ®broblasts, including telomeric associations, dicentrics, ring chromosomes, breaks, chromatid exchanges and translocations. These aberrations may be divided into two groups depending on their putative origins: the ®rst group may be linked to telomere erosion and includes telomere associations and dicentrics; the second group may be linked to inactivation of G2 checkpoint function and includes breaks, exchanges, translocations and ring chromosomes. Both groups of aberrations developed progressively as F5E6 cells aged to crisis, while the ®rst group appeared to be diminished post-crisis (Table 1). One of the functions of telomeres is to prevent endto-end associations and fusions between chromosomes. When two chromosomes lose telomeric DNA they may associate at their termini as a ®rst step that precedes formation of a true dicentric chromosome. When the centromeres of the dicentric chromosome segregate towards the two daughter poles during anaphase, a chromosome break may occur between the two centromeres. The new chromosome ends are non-telomeric and if carried through another replication cycle may again fuse to form another dicentric. A cycle of chromatid bridge-breakage and fusion was originally described by McClintock for maize chromosomes (McClintock, 1941; see Coquelle et al., 1997 for recent discussion) but without the ®rst apparent step of forming chromosome end-to-end associations, and this model has been invoked to explain the mechanisms of gene ampli®cation in cells incubated with DNA metabolic poisons (Coquelle et al., 1997). The number of telomere associations and dicentrics was signi®cantly increased with the aging of F5E6 ®broblasts, consistent with a view that progressive telomere erosion in cells that lack p53 and G1 checkpoint function leads to the formation of these structures. A connection between telomere associations/dicentrics and telomere erosion is also suggested by the observation that in postcrisis F5E6 cells, which expressed telomerase to stabilize telomeres, the frequency of cells with these lesions was signi®cantly reduced. Interestingly, ataxia telangiectasia (AT) cell lines that are defective in G1 and G2 checkpoint functions (Kaufmann and Paules, 1996; Rotman and Shiloh, 1997), displayed increased frequencies of chromosome end-to-end-associations (Pandita et al.,


1995, 1996). An inverse correlation between telomere length and chromosome end associations was observed in AT cells (Pandita et al., 1995). Another origin of dicentric chromosomes as well as chromosome translocations and deletions are DNA double-strand breaks occuring during G1 phase in the non-telomeric regions of chromosomes. These may occur at fragile sites (Yunis and Soreng, 1984; Coquelle et al., 1997), constitutive heterochromatin regions, and in chromosomal regions (mostly R-bands) containing clusters of Alu-family repeats (Filatov et al., 1987, 1991); Korenberg and Rykowski, 1988). AT cells also expressed higher frequencies of chromosomal breaks at metaphase (Pandita et al., 1995). The second group of chromosomal aberrations found in F5E6 cells includes chromosome breaks, translocations, chromatid breaks, chromatid exchanges and ring chromosomes. The chromatid-type aberrations are especially interesting as these originate during S phase from lesions in only one of the parental DNA strands or from double-strand breaks introduced into late S and G2 phase cells. The appearance of chromatid-type aberrations in F5E6 metaphases implies some defect in DNA damage response or DNA repair. As chromosomal aberrations were highly correlated with inactivation of G2 checkpoint function, which serves to prevent cells from entering mitosis with broken chromosomes, it is possible that loss of this checkpoint permits cells to enter mitosis with damage that would normally arrest growth. G2 checkpoint function remained attenuated in post-crisis F5E6 cells that express telomerase and this may account for the persistent chromosome and chromatid aberrations. Spontaneously immortalized LFS fibroblasts and SV40-immortalized ®broblasts also display attenuation of G2 checkpoint function indicating that the defect in this checkpoint post-crisis is independent of telomerase and stabilization of telomeres (Kaufmann et al., 1995; Paules et al., 1995). The ring chromosomes seen in this series may have two dierent origins. Usually ring chromosomes are accompanied by double fragments. In such cases the origin of the ring chromosome is a result of doublestrand breaks in both chromosome arms followed by fusion of these newly formed chromosome ends. Another way to form a ring chromosome is by telomere shortening. The ring chromosome shown in Figure 3b did not have an accompanying double fragment and may be formed by this mechanism. We found an involvement of chromosomes 4 and 6 in translocations with other chromosomes in F5E6 cells at passage 27, which is consistent with the literature on chromosomes involved in senescence and aging (Smith and Pereira-Smith, 1996; Ning et al., 1991; SolinasToldo et al., 1997). One of the chromosome 4 rearrangements was an isodicentric chromosome with breakpoints located very close to the telomere on the q-arm. This idodicentric could have been formed by telomere shortening and chromatid fusion. E6-expressing cells also displayed highly signi®cant changes in chromosome numbers and ploidy suggestive of non-disjunction errors and mitotic spindle abnormalities (Fukasawa et al., 1996). However, signi®cant ploidy changes did not occur until there had been substantial telomere erosion and attenuation of G2 checkpoint function. A recent analysis of the mechan-

ism of chromosomal endoreduplication in murine cells indicated that p53 acts to prevent initiation of DNA synthesis in cells that pass from mitotic arrest directly into G1, without having completed the anaphase and telophase steps of mitosis (Minn et al., 1996). In normal human ®broblasts, the p53-dependent mitotic spindle checkpoint exercises stringent control over initiation of DNA synthesis. As shown in Figure 4, less than 2% of normal ®broblasts endoreduplicate DNA after 24 h incubation with colcemid. The nature of the signal that induces p53 to arrest S phase entry in colcemid-treated cells has not been determined. It is notable that polyploidization of E6-expressing ®broblasts was also correlated with chromosomal aberrations and attenuation of G2 checkpoint function. Although the E6 cells displayed the ability to endoreduplicate when stressed with the mitotic poison colcemid, they maintained a diploid karyotype at least through passage 16. This result raises the possibility that some element of genetic destabilization associated with chromosomal damage or attenuation of G2 checkpoint produces a stress on the mitotic spindle checkpoint system, thereby selecting for outgrowth of polyploid cells. The absence of numerical and structural chromosome alterations in p53-negative E6-expressing cells at low passages shows that there is no simple relationship between p53 loss or mutation and chromosomal instability (Yin et al., 1992; Livingstone et al., 1992; White et al., 1994; Lengauer et al., 1997). In our study, low passage p53-negative E6 cells with attenuated mitotic spindle checkpoint function expressed a functional G2 checkpoint and displayed a normal telomere length distribution. Only in aged cells with attenuated G2 checkpoint and eroded telomeres was chromosomal destabilization manifested. The results suggest that chromosomal destabilization in telomerase-negative cells that express HPV16 E6 may involve selective outgrowth of G2 checkpoint-defective cells during a phase of telomere crisis. Expression of telomerase post-crisis was associated with incomplete normalization of chromosomal numbers and structure suggestive of persistent chromosomal instability. In summary, the simultaneous analysis of checkpoint functions and telomere structure provided new insights to the mechanisms of chromosome destabilization in human cells. The results indicate that there is a complex interplay between telomere erosion and attenuation of G2 checkpoint function that accompanies the instabilities for both chromosome numbers and structure. The observation that chromosomal instability was associated with both telomere erosion and inactivation of G2 checkpoint function precludes a conclusion as to which is the rate-limiting step. Targeted inactivation of G2 checkpoint function in low passage E6-expressing cells with long telomeres may enable determination of the rate-limiting step in chromosome destabilization. Materials and methods Cells The diploid human ®broblast line, NHF5, was derived from foreskin and grown in minimal essential medium supplemented with 10% fetal bovine serum, L-glutamine

1835


1836

and antibiotics as previously described (Paules et al., 1995; Kaufmann et al., 1997). Primary cultures of NHF5 cells were infected with a replication-defective amphotropic retrovirus containing a neomycin-resistance gene and designated, F5neo, or with the retrovirus containing the HPV-16E6 oncogene, and designated, F5E6 (Kaufmann et al., 1997). Infected cells were selected by growth in G418. F5E6 and isogenic control F5neo cells were provided by Dr Denise Galloway (Fred Hutchinson Cancer Research Center). These cells were passaged every week using a 1 : 8 split ratio. At various passage levels during in vitro population expansion, aliquots of cells were cryopreserved. After secondary cultures of F5neo cells underwent senescence arrest, cultures of F5neo and F5E6 cells were reconstituted from aliquots frozen at passage levels 4, 15, 21 and 26. Re-constituted lines were used generally within two passages. F5E6 cells that had been in culture continuously up to passage 34 were also included in some analyses. Mitotic spindle and G2 checkpoint assays The dependence of initiation of DNA synthesis on completion of mitosis was assayed by incubating logphase cells with 0.27 mM colcemid for 24 h. During the ®nal 2 h of incubation BrdU was added at 10 mM. Fixed cells were analysed for DNA content and incorporation of BrdU by ¯ow cytometry as previously described (Kaufmann et al., 1997). G2 delay was assayed by ¯uorescence microscopy, also as described in Kaufmann et al. (1995). Log-phase cells were irradiated with 1.5 Gy or shamtreated (as controls) then incubated for 2 h at 378C before ®xation with methanol/acetic acid (3 : 1). After staining with propidium iodide, mitotic cells were enumerated by ¯uorescence microscopy and expressed as a percentage of the total (mitotic index). A minimum of 2000 cells were counted for each sample. To ensure an unbiased analysis, samples were assigned coded random numbers before quantitation of mitotic index. Cytogenetics Fibroblasts were split 1 : 4 1 day before harvesting when colcemid (0.05 mg/ml) was added for 25 min to arrest cells in metaphase. Cells were detached with trypsin/EDTA, incubated in 0.075 M KCl for 20 min at 378C, and ®xed in three changes of methanol:acetic acid (3 : 1). Fixed cell pellets were used for slide preparation. Slides were air dried at least 2 days at 378C before using. For chromosome aberrations and polyploidy study, slides were stained with Giemsa solution (Gibco, BRL) in 0.06 M phosphate buer (pH 6.8, 1 : 30) for 4 min at room temperature. At each passage level 50 metaphases were examined for quantitation of chromosome aberrations and chromosome numbers, and 500 metaphases were analysed for ploidy, except for passage 34 F5E6 cells, in which 33 metaphases were examined for cytogenetics and 50 metaphases for determination of ploidy. Fluorescence in situ hybridization (FISH) Probes The probes used for chromosome painting were Spectrum Green WCP 4 (for chromosome 4) directly labeled with ¯uorescein-12-dUTP and Spectrum Orange WCP 6 (for chromosome 6) directly labeled with rhodamine (Vysis, Inc.). For analysis of telomere erosion an all-human telomeric probe labeled with digoxigenin (Oncor, Inc.) was used. Slide-preparation Air-dried slides were aged two to three weeks prior to FISH, denatured in 70% formamide/ 26SSC, pH 7.0 at 728C for 2.5 min and gradually dehydrated in cold ethanol series (70%, 85%, 100%).

Probe preparation, hybridization, postwashing and detection The slides from the cells at passage 27 were used for FISH with Spectrum Orange chromosome 6 and Spectrum Green chromosome 4 probes. Two microliters of Spectrum Green WCP 4 and 2 ml of Spectrum Orange WCP 6 probes were mixed with 2 ml of water and 14 ml of hybridization buer (50% formamide/10% dextran-sulfate/26SSC). WCP 4 and WCP 6 DNA was denatured 5 min at 738C. The denatured probe was added to slides, sealed under glass coverslip with rubber cement and incubated overnight at 378C. Postwashing was performed at 458C in three changes of 50% formamide/26SSC, pH 7.0 for 5 min each, in 26SSC, pH 7.0 for 5 min and ®nally at 26SSC/ 0.1% NP-40 for 5 min (Filatov et al., 1996). Slides were counterstained with DAPI (4'-6-diamidino-2-phenildole) in Antifade solution (18 ml per slide; Vysis, Inc.). FISH with telomeric DNA probe The slides from F5-neo cells at passages 5, 16, 22, 27 and F5-E6 cells at passages 5, 16, 22, 27 and 34 were used for FISH with an `all-human-telomeres' probe (Oncor, Inc.). Digoxigenin-labeled telomeric probe was hybridized and detected according to the supplier's protocol: 30 ml of probe were denatured for 5 min at 738C, added to slides, and incubated overnight at 378C. The hybridization signals were detected by incubation with ¯uorescein-labeled anti-digoxigenin antibody for 5 min at 378C, followed by amplification with rabbit anti-sheep and ¯uorescein-labeled antirabbit antibodies for 15 min each. The chromosomes were counterstained with propidium iodide (0.03 mg/ml; Oncor, Inc.). Cells were viewed and photographed with a Zeiss photomicroscope (Axiophot 20) equipped for epifluorescence. Kodak Ektachrome color slide 400 ®lm was used for photography. Telomere spot numbers were determined for 10 diploid metaphases per sample. A photomicroscopic method for quanti®cation of telomere signal intensity over diploid metaphases was also employed (Lansdorp et al., 1996). Exposure time (in s), that is inversely proportional to intensity, was automatically measured by the Zeiss photomicroscope exposure system. Telomere terminal restriction fragment (TRF) analysis Genomic DNA was extracted from cells by a standard protocol. Cell pellets were resuspended in 1 vol of lysis buer, containing 10 mM Tris-HCl, pH 8.0, 100 mM NaCl, 25 mM EDTA, pH 8.0, 0.5% sodium dodecyl sulfate, 0.1 mg/ml proteinase K. The samples were incubated at 508C for 12 ± 16 h, extracted with phenol/chloroform/ isoamyl alcohol and precipitated with ethanol. The DNA was dissolved in Tris-EDTA buer. RNA was removed by adding DNase free-RNase (1 mg/ml) and incubating for 1 h at 378C. Genomic DNA was digested with the restriction enzymes HinfI and RsaI for 16 h at 378C. The digested DNA was extracted with phenol-chloroform, ethanol precipitated and dissolved in Tris-EDTA buer. DNA was electrophoresed in an 0.8% agarose gel in Tris-borateEDTA buer for 16 h. Southern hybridization was performed with a non-radioactive digoxigenin-labeled (TTAGGG)3 probe at 438C for 18 h using the Genius System protocol (Boehringer Mannheim). Telomerase assay Cell extracts for assay of telomerase were prepared as described (Golubovskaya et al., 1997). The concentration of protein was measured for each extract with bicinchroninic acid protein assay kit (Pierce Chemical Co). Telomerase activity was detected by the PCR-mediated, telomeric-repeat ampli®cation protocol (TRAP) described by Kim et al., 1994). The initial telomerase extension


reaction and the secondary PCR reaction were performed in dierent tubes (Bryan et al., 1995; Nakayama et al., 1997), as described (Golubovskaya et al., 1997).

5 mM potassium ferrocyanide, 150 mM NaCl and 2 mM MgCl2.

b-galactosidase assay

Acknowledgements We would like to thank Dr JM Mason for his critical reading of this manuscript and helpful comments and discussion. We are grateful to members from Medical Illustrations and Photography Department for expert assistance in preparation of the ®gures. This work was supported by Public Health Service grant CA42765 (WKK).

Senescent ®broblasts were identi®ed by assay of pH 6.0 bgalactosidase as described by Dimri et al. (1995). Brie¯y, cells were washed in PBS, ®xed for 20 min in 2% formaldehyde, 0.2% glutaraldehyde, washed again in PBS and incubated for several hours in staining solution: 1 mg/ ml X-Gal (5-bromo-4-chloro-3-indolyl b-D-galactoside), 40 mM citric acid, 40 mM sodium phosphate, pH 6.0, References Blasco MA, Lee H-W, Hande MP, Samper E, Landsorp PM, DePinho RA and Greider CW. (1997). Cell, 91, 25 ± 34. Bond JA, Wyllie FS and Wynford-Thomas D. (1994). Oncogene, 9, 1885 ± 1889. Bryan TM, Englezou A, Gupta J, Bacchetti S and Reddel RR. (1995). EMBO J., 14, 4240 ± 4248. Chang THT, Ray FA, Thompson DA and Schlegel R. (1997). Oncogene, 14, 2383 ± 2393. Cohen MM and Levy HP. (1989). Adv. Human Genet., 18, 43 ± 149. Counter CM, Avilion AA, LeFeuvre CE, Stewart NG, Greider CW, Harley CB and Bacchetti S. (1992). EMBO J., 11, 1921 ± 1929. Coquelle A, Pipiras E, Toledo F, Buttin G and Debatisse M. (1997). Cell, 89, 215 ± 225. Cross SM, Sanchez CA, Morgan CA, Schimke MK, Ramel S, Idzerda RL, Raskind WH and Reid BJ. (1995). Science, 267, 1353 ± 1356. DiLeonardo A, Khan SH, Linke SP, Greco V, Seidita G and Wahl GM. (1997). Cancer Res., 57, 1013 ± 1019. Dimri GP, Lee X, Basile G, Acosta M, Scott G, Roskelley C, Medrano EA, Linskens M, Rubelj I, Pereira-Smith O, Peacocke M and Campisi J. (1995). Proc. Natl. Acad. Sci. USA., 92, 9363 ± 9367. Dulic V, Kaufmann WK, Wilson SJ, Tlsty TD, Lees E, Harper JW, Elledge SJ and Reed SI. (1994). Cell, 76, 1013 ± 1023. Ejima Y and Sasaki MS. (1986). Mutat Res., 159, 117 ± 123. Filatov LV, Mamaeva SE and Tomilin NV. (1987). Mol. Biol. Rep., 12, 117 ± 121. Filatov LV, Mamaeva SE and Tomilin NV. (1991). Cancer Genet Cytogenet., 56, 11 ± 22. Filatov LV, Saito M, Behm FG, Rivera GK and Raimondi SC. (1996). Cancer Genet. Cytogenet., 89, 136 ± 140. Fukasawa K, Choi T, Kuriyama R, Rulong S and Vande Woude GF. (1996). Science, 271, 1744 ± 1747. Galloway DA, Demers GW, Foster SA, Halbert CL and Russell K. (1994). Cold Spring Harbor Symp. Quant. Biol., LIX, 297 ± 306. Golubovskaya VM, Presnell SC, Hooth MJ, Smith GJ and Kaufmann WK. (1997). Oncogene, 15, 1233 ± 1240. Grana X and Reddy P. (1995). Oncogene, 11, 212 ± 219. Greider CW and Blackburn EH. (1985). Cell, 43, 405 ± 413. Harley CB, Futcher AB and Greider CW. (1990). Nature, 345, 458 ± 460. Hartwell LH. (1992). Cell, 71, 543 ± 546. Hartwell LH and Kastan MB. (1994). Science, 266, 1821 ± 1828. Healy KM. (1995). Oncology Res., 7, 121 ± 130. Henderson SR, Spector D, Wang S-S and Harley C. (1996). J. Cell. Biol., 134, 1 ± 12. Holliday R. (1989). Trends in Genetics, 5, 42 ± 44. Ishikawa F. (1997). Bio. Biophys. Res. Commun., 230, 1 ± 6. Kaufman WK, Levedakou EN, Grady HL, Paules RS and Stein GH. (1995). Cancer Res., 55, 7 ± 11.

Kaufmann WK and Kaufman DG. (1993). FASEB J., 7, 1188 ± 1191. Kaufmann WK and Paules RS. (1996). FASEB J., 10, 238 ± 247. Kaufmann WK, Schwartz JL, Hurt JC, Byrd LL, Galloway DA, Levedakou E and Paules RS. (1997). Cell Growth Dier., 8, 1105 ± 1114. Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, Ho PLC, Coviello GM, Wright WE, Weinrich SL and Shay JW. (1994). Science, 266, 2011 ± 2015. Klingelhutz AJ, Barber SA, Smith PP, Dyer K and McDougall JK. (1994). Mol. Cell. Biol., 14, 961 ± 969. Korenberg JR and Rykowski MC. (1988). Cell, 53, 391 ± 400. Lansdorp PM, Verwoerd NP, van de Rijke FM, Dragovska V, Little M-T, Dirks RW, Raap AK and Tanke HJ. (1996). Human Molecular Genetics, 5, 685 ± 691. Lengauer C, Kinzler KW and Vogelstein B. (1997). Nature, 386, 623 ± 626. Levedakou EN, Kaufmann WK, Alcorta DA, Galloway DA and Paules RS. (1995). Cancer Res., 55, 2500 ± 2502. Livingstone LR, White A, Sprouse J, Livanos E, Jacks T and Tlsty TD. (1992). Cell, 70, 923 ± 935. McClintock B. (1941). Genetics, 26, 234 ± 282. Minn AJ, Boise LH and Thompson CB. (1996). Genes Develop., 10, 2621 ± 2631. Nakayama J, Saito M, Nakamura H, Matsuura A and Ishikawa F. (1997). Cell, 88, 875 ± 884. Ning Y, Weber JL, Killary AM, Ledbetter DH, Smith JR and Pereira-Smith OM. (1991). Proc. Natl. Acad. Sci. USA, 88, 5635 ± 5639. Olovnikov AM. (1971). Doklady Biochem., 201, 394 ± 397. Pandita TK, Pathak S and Geard CR. (1995). Cytogenet. Cell Genet., 71, 86 ± 93. Pandita TK, Hall EJ, Hei TK, Piatyszek MA, Wright WE, Piao CQ, Pandita RK, Willey JC, Geard CR, Kastan MB and Shay JW. (1996). Oncogene, 13, 1423 ± 1430. Paules RS, Levedakou EN, Wilson SJ, Innes CL, Rhodes N, Tlsty TD, Galloway DA, Donehower LA, Tainsky MA and Kaufmann WK. (1995). Cancer Res., 55, 1763 ± 1773. Rezniko CA, Belair C, Savalieva E, Zhai Y, Pfeifer K, Yeager T, Thompson KJ, DeVries S, Bindley C, Newton MA, Sekhon G and Waldman F. (1994). Genes Devel., 8, 2227 ± 2240. Rogan EM, Bryan TM, Hukku B, Maclean K, Chang ACM, Moy EL, Englezou A, Warneford SG, Dalla-Pozza L and Reddel RR. (1995). Mol. Cell. Biol., 15, 4745 ± 4753. Rotman G and Shiloh Y. (1997). BioAssays, 19, 911 ± 917. Schener M, Werness BA, Huibregtse JM, Levine AJ and Howley PM. (1990). Cell, 63, 1129 ± 1136. Schwartz JL, Cowan J, Grdina DJ and Weichselbaum RR. (1996). Radiat Res., 146, 139 ± 143. Shay JW, Wright WE, Brasiskyte D and Van Der Haegen BA. (1993). Oncogene, 8, 1407 ± 1413. Smith JR and Pereira-Smith OM. (1996). Science, 273, 63 ± 67.

1837


1838

Solinas-Toldo S, Durst M and Lichter P. (1997). Proc. Natl. Acad. Sci. USA, 94, 3854 ± 3859. Stewart BG and Bacchetti S. (1991). Virology, 180, 49 ± 57. Taylor AMR, Metcalfe JA, Oxford JM, and Harnden DG. (1976). Nature, 260, 441 ± 443. Therkelsen AJ, Nielsen A, Koch J, Hindkjaer J and Kolvraa S. (1995). Cytogenet. Cell Genet., 68, 115 ± 118. Van Steensel B and de Lange T. (1997). Nature, 385, 740 ± 743.

White AE, Livanos EM and Tlsty TD. (1994). Genes Devel., 8, 666 ± 677. Yin Y, Tainsky MA, Bischo FZ, Strong LC and Wahl G. (1992). Cell, 70, 937 ± 948. Yunis JJ and Soreng AL. (1984). Science, 226, 1199 ± 1204. Zampetti-Bosseler F and Scott D. (1981). Inter. J. Radiat Biol., 39, 547 ± 558.