Alternative Lengthening of Telomeres Is Associated with Chromosomal Instability in Osteosarcomas

Alternative lengthening of telomeres is associated with chromosomal instability in osteosarcomas

Christina Scheel1, Karl-Ludwig Schaefer1, Anna Jauch3, Monika Keller3, Daniel Wai1, Christian Brinkschmidt1, Frans van Valen2, Werner Boecker1, Barbara Dockhorn-Dworniczak1 and Christopher Poremba*,1

1Gerhard-Domagk-Institute of Pathology, WestfaÈlische Wilhelms University, MuÈnster, Germany; 2Department of Experimental Orthopedics, WestfaÈlische Wilhelms University, MuÈnster, Germany; 3Department of Human Genetics, Ruprecht Karls University, Heidelberg, Germany

Telomere maintenance is regarded as a key mechanism in most cases this is achieved by the activation of in overcoming cellular senescence in tumor cells and in telomerase (Greider and Blackburn, 1987; Counter et most cases is achieved by the activation of telomerase. al., 1998). Earlier investigations revealed telomerase However there is at least one alternative mechanism of activity (TA) in 90% of human tumors, but not in telomere lengthening (ALT) which is characterized by normal tissues (Kim et al., 1994; Shay and Bacchetti, heterogeneous and elongated telomeres in the absence of 1997). telomerase activity (TA). We evaluated the prevalence of However, the speci®c association of telomerase with TA, gene expression of telomerase subunits and ALT in cancer progression is being challenged. The frequency relation to telomere morphology and function in matrix and biologic importance of TA seem to be subjected to producing bone tumors and in osteosarcoma cell lines tumor-type dependent variation (Wynford-Thomas, and present evidence of a direct association of ALT with 1999). Stem cells (Harle and Boukamp, 1996), telomere dysfunction and chromosomal instability. stimulated lymphocytes (Broccoli et al., 1995; Hiyama Telomere ¯uorescence in situ hybridization (T-FISH) in et al., 1995; Counter et al., 1995) and dierent highly ALT cells revealed elongated and shortened telomeres, proliferative tissues (Hiyama et al., 1996; Takubo et al., partly in unusual con®gurations and loci, dicentric 1997; Bryan et al., 1997) exhibit TA. Moreover, an marker chromosomes and signal-free chromosome ends. alternative mechanism of telomere lengthening (ALT) Free ends give rise to end-to-end associations and may exists in a subset of human tumors and immortalized induce breakage-fusion-bridge cycles resulting in an cell lines. ALT is characterized by the presence of increased number of complex chromosomal rearrange- elongated and heterogeneous telomeres in the absence ments, as detected by multiplex-FISH (M-FISH). We of TA and is supposed, though not yet elucidated, to propose that ALT cannot be seen as an equivalent to elongate telomeres through recombination-based inter- telomerase activity in telomere maintenance. Its associa- chromosomal exchanges of sequential material (Bryan tion with telomere dysfunction and chromosomal in- et al., 1995; Biessmann and Mason, 1997; Shay and stability may have major implications for tumor Bacchetti, 1997; Bryan et al., 1997). progression. Oncogene (2001) 20, 3835 ± 3844. Our purpose was to evaluate the prevalence of TA and ALT in matrix producing bone tumors in regard Keywords: telomere maintenance; osteosarcomas; telo- to telomere morphology and function. We ®rst laid merase activity; alternative lengthening of telomeres; emphasis on the analysis of the frequency of TA in chromosomal instability osteo- and chondrosarcomas, benign lesions with various histology, and osteosarcoma cell lines by the telomere repeat ampli®cation protocol (TRAP). Gene Introduction expression of the enzyme's catalytic subunit hTERT (Meyerson et al., 1997; Takakura et al., 1998; Telomeres shorten with each round of cell division Ramakrishnan et al., 1998) and its endogenous RNA (Olovnikov, 1971; Watson, 1972). Telomere mainte- template hTR (Avilion et al., 1996; Ito et al., 1998; nance is regarded as a key mechanism in overcoming Sumida et al., 1999) was performed by quantitative cellular senescence through immortalization in tumor real-time RT ± PCR. Secondly, we detected ALT in cells (Wright et al., 1989; Counter et al., 1992) and primary osteosarcomas and osteosarcoma cell lines by telomere length analysis using Southern blotting. Analysis of telomere morphology and function was performed by telomere PNA-¯uorescence in situ *Correspondence: C Poremba, Gerhard-Domagk-Institute of hybridization (T-FISH). Finally we investigated the Pathology, WestfaÈ lische Wilhelms-University, Domagkstrasse 17, occurrence of molecular cytogenetic aberrations char- 48149 MuÈ nster, Germany Received 4 January 2001; revised 27 March 2001; accepted 2 April acteristic for either TA or ALT by comparative 2001 genomic hybridization (CGH) in osteosarcomas and ALT is associated with chromosomal instability C Scheel et al 3836 osteosarcoma cell lines and by 24-color multiplex FISH expression (Table 3). The osteosarcoma cell lines U2- (M-FISH) in osteosarcoma cell lines. As a guardian of OS and ZK-58 had no detectable TA in the TRAP genome integrity, TP53 gene status in the osteosarco- assay and no hTERT expression. ZK-58 presented ma cell lines was analysed. marginal hTR expression and U2-OS had no detectable hTR expression. Neuroblastoma cell lines Lan-5 and SK-PN-DW, used as positive controls in T-FISH and telomere length analysis, showed high levels of TA. To Results exclude false negative results, protein extracts of telomerase positive tumors were mixed with protein Frequency of TA in matrix producing bone tumors extracts of telomerase negative tumors at dierent Sixty-eight bone tumors were assayed for TA by the dilutions. No inhibition of the PCR reactions of the TRAP assay (Table 1). 15/29 osteosarcomas (52%) TRAP assay was detected. compared to 16/23 chondrosarcomas (70%) showed TA. 6/16 histologically dierent benign lesions (38%), ALT, TA and telomere morphology all of chondrogenic origin, presented minimal TA. TA levels were signi®cantly lower than in any of the The 4/7 cell lines with high TA (HOS, MNNG, OST, examined malignant tumors. When examining the SJSA-1) had short telomeres, with a mean length of corresponding histological slides, bone marrow cells 3.5 kb (Figure 1). Cell lines U2-OS and ZK-58 and in®ltration of lymphocytes in some regions of these displayed no TA and no hTERT expression. Telomere lesions were found. Contamination with these cells in length analysis of these cell lines by Southern blotting the corresponding samples could explain our ®ndings revealed a hybridization signal smear stretching from of minimal TA (Counter et al., 1995; Harle and 52 ± 23 kb with the strongest hybridization signal Boukamp, 1996). present at 23 kb, thus representing the ALT phenotype 15/68 primary tumors (eight osteosarcomas, seven (Bryan et al., 1997). Saos-2 exhibited an intermediate chondrosarcomas) were assayed for telomerase core ALT type with low TA, marginal hTERT and hTR component (hTERT, hTR) gene expression by quanti- expression, but elongated and heterogeneous telomeres tative real-time RT ± PCR (Table 2). 8/8 tumors with (strongest hybridization signal at 23 kb). TA (100%) were positive for hTERT gene expression To exclude that ALT is a cell culture artifact, DNA and 7/7 tumors lacking TA (100%) had no detectable of two primary high grade osteoblastic osteosarcomas hTERT expression. All tumors revealed hTR expres- obtained from tumor biopsies without any pretreat- sion with a tendency to higher expression in telomerase ment was included for telomere length analysis (Figure positive tumors, consistent with the literature (Avilion 1). One tumor without TA and without hTERT et al., 1996; Ito et al., 1998). expression revealed elongated and heterogeneous Cell lines MNNG, HOS, OST and SJSA-1 displayed telomeres with a mean length of 23 kb, representing high TA with varying, but signi®cantly higher levels of the ALT phenotype. The other tumor, with inter- hTERT and hTR gene expression when compared to mediate TA and high hTERT expression, had Saos-2, which exhibited low TA and hTERT/hTR gene telomeres with a mean length of 4 kb.

Fluorescence in situ hybridization with a telomere specific Table 1 Frequency of telomerase activity in matrix producing bone PNA-probe (T-FISH) tumours Twenty metaphases each of neuroblastoma cell line Entity Number Telomerase Activity % Lan-5 and osteosarcoma cell line OST (both with high Total number 68 37 54 TA), ALT cell lines U2-OS and ZK-58, intermediate Malignant tumors 52 31 60 type cell line Saos-2 and a normal human lymphocyte Osteosarcoma 29 15 52 high grade 25 13 52 culture were examined. Aberrations present in at least low grade 2 0 0 50% (10 metaphases) are described. Chromosomes of post chemo 4 1 25 Lan-5, OST or the human normal lymphocyte culture relapse 2 0 0 essentially showed four ¯uorescent hybridization not further class. 2 1 50 signals at telomeric positions of similar ¯uorescent Chondrosarcoma 23 16 70 GI 6 5 83 intensity (Figure 2a). In Saos-2, U2-OS and ZK-58, GII 13 7 54 ¯uorescence intensity diered greatly in chromosomes GIII 4 4 100 of the same metaphase and between chromosome ends. Benign lesions*166*38Hybridization spots with a higher and a lower intensity En/Chondroma 7 5 71 Eosinophilic granuloma 3 0 0 than in Lan-5, OST and the lymphocyte culture were Giant cell tumor 1 0 0 found (Figure 2b). Desmoplastic fibroma 1 0 0 Large, dicentric marker chromosomes indicating Aneurysmatic bone cyst 2 0 0 end-to-end fusion were present. Telomeric repeats were Chondrogenic metaplasia 1 1 100 found at unusual loci and con®gurations. Some Synovial chondromatosis 1 0 0 chromosomes had no detectable telomeric repeats at *Low telomerase activity (+) one or both ends (signal-free ends), others displayed

Oncogene ALT is associated with chromosomal instability C Scheel et al 3837 from one up to four signals at chromosome ends. 24-color multiplex fluorescence in situ hybridization Telomeric signals at ectopic and intrachromosomal (M-FISH) centromeric positions were detected (Figure 2c ± e). Among chromosomes with aberrant telomere mor- Metaphases of three osteosarcoma cell lines (ZK-58 phology, chromosomes with a `normal' telomeric n=10, Saos-2 n=5, MNNG n=3) were analysed for phenotype were found ruling out the possibility of structural (=rearrangements) and numerical (=aneu- failed or unspeci®c hybridization (Figure 2b). All slides ploidy) chromosomal aberrations. In the osteosarcoma were hybridized in the same assay and the procedure cell line Saos-2 with a modal number of 52 chromo- was performed twice with consistent results. somes, 37 chromosomes (71.2%) were aected by structural aberrations, whereby a high rate of complex rearrangements composed of 3 ± 5 dierent chromo- Comparative genomic hybridization (CGH) somes was found (12/37) (Figures 3 and 4b). Very The total number of molecular cytogenetic aberrations similar results were achieved by M-FISH analysis in in the 24 analysed osteosarcomas ranged from three up the cell line ZK-58. Structural abnormalities were to 35 aberrations with a mean total number of detected in 69.8% (37/53) of the chromosomes, where- aberrations of 19+9. The seven osteosarcoma cell by 10 chromosomes showed complex rearrangements lines analysed revealed a mean total number of 29+7 composed of 3 ± 5 dierent chromosomal segments aberrations ranging from 18 ± 33 aberrations per cell (Figure 3). In contrast, in the osteosarcoma cell line line. However, number and distribution of gains and losses of genetic material was heterogeneous and no loci were shared by all tumors and/or cell lines, or by tumors grouped according to their telomerase status. Emphasis was laid on the analysis of chromosomal regions which are related to the regulation of TA. There is evidence for a cellular telomerase repressor at chromosome 3p (Tanaka et al., 1998; Shay, 1999). The hTERT gene was recently mapped to chromosome 5p15.33 by ¯uorescence in situ hybridization (Bryce et al., 2000). Tumors were grouped into 13 tumors with and 11 tumors without TA. The Chi-square test and Fisher's Exact test for the parameters of gains/ ampli®cations and losses on chromosome 3p and 5p, respectively, revealed no dierences between both groups.

Table 2 Correlation of telomerase activity and hTERT gene expression in malignant bone tumors hTERT+/TA+ hTERT+ hTERT7 Entity n hTERT7/TA7 Figure 1 Telomere length analysis. A neuroblastoma cell line, SK-PN-DW, with high levels of telomerase activity (TA) is used TA+ 3 0 Osteosarcoma 8 8 (100%) as a positive control for all osteosarcoma cell lines included in this TA7 05 study. Furthermore two high grade, highly cellular primary TA+ 5 0 Chondrosarcoma 7 7 (100%) osteosarcomas (indicated by Tu) are included, one without TA TA7 02 exhibiting the ALT phenotype and the other with TA and short TA+ 8 0 Sum 15 15 (100%) telomeres. TeloLOW and TeloHIGH are Hinf1/Rsa1 digested TA7 07 control DNAs with a mean length of 3.3 and 11.3 kb, respectively

Table 3 Telomere dynamics in osteosarcoma cell lines Cell lines TA hTERT hTR Mean telomeres (kb) CGH TP53 status

MNNG +++ 51.5 9.9 3.0 33 Point mutation Codon 156 (Radig et al., 1998) HOS +++ 86.2 25.5 3.5 18 Point mutation Codon 152 (Romano et al., 1989) OST +++ 28.3 100.0 3.5 37 Wild type+ SJSA-1 +++ 100.0 42.5 3.5 23 Wild type+ Saos-2 + 0.1 0.01 23.0 33 large deletion (Chen et al., 1990) U2-OS 0.0 0.0 0.0 23.0 33 Wild type, 7 fold MDM2 amplification (Landers et al., 1997) ZK-58 0.0 0.0 0.01 23.0 29 Wild type+

+determined using the TP53 Gene Chip (Aymetrix). TA by TRAP: telomerase activity assayed by the TRAP assay. hTERT, hTR: hTERT and hTR expression normalized on GAPDH expression; values relative to highest relative gene expression which was assigned relative value 100 (%). Mean telomeres: mean telomere length in kilobases (kb). CGH: total number of molecular cytogenetic aberrations assayed by Comparative Genomic Hybridization (CGH)

Oncogene ALT is associated with chromosomal instability C Scheel et al 3838

Figure 2 Telomere PNA-FISH. (a) metaphase of Lan-5, a telomerase positive neuroblastoma cell line used as a positive control. (b) metaphase of intermediate ALT-type osteosarcoma cell line Saos-2. Representative metaphase chromosomes of cell lines (c) Saos-2, (d) U2-OS and (e) ZK-58

Figure 3 M-FISH diagram. Shown are the modal chromosome number, the total number of rearrangements subdivided into translocations and deletions, complex rearrangements (with three up to ®ve dierent chromosomes involved) and trisomies for osteosarcoma cell lines Saos-2 (intermediate ALT-type), ZK-58 (ALT) and MNNG (TA)

Oncogene ALT is associated with chromosomal instability C Scheel et al 3839

Figure 4 M-FISH metaphases. Metaphase of osteosarcoma cell lines (a) MNNG (TA) and (b) Saos-2 (intermediate ALT-type). (c) Details of complex rearrangements in Saos-2 for chromosomes 1, 10, 11. (d) Chromosomal composition of complex rearranged chromosomes 2 and 1 in Saos-2

MNNG, M-FISH detected structural aberrations only ably, the histologic tumor grade of the osteosarcomas in 29.9% (18/60) of the chromosomes, whereby three was dierent to that of the chondrosarcomas: almost chromosomes consisted of three dierent chromosomal all osteosarcomas included in the study were high segments (Figures 3 and 4a). Additionally, M-FISH grade, highly cellular tumors. In contrast, the identi®ed a trisomy for chromosomes 1, 2, 5 and 22. collective of 23 chondrosarcomas comprised only When comparing ZK-58 (no TA, elongated telomeres) four high grade, highly cellular chondrosarcomas, and Saos-2 (low TA, elongated telomeres) with while six were low grade and 13 were intermediate MNNG (high TA, short telomeres) using the Wilcoxon grade, matrix-rich tumors with low to moderate test the proportion of complex rearranged chromo- cellularity. All high grade chondrosarcomas were somes (with 3 ± 6 chromosomes involved; Figure 4c,d) positive for TA. The failure to detect TA in some as well as the overall number of rearrangements was of the low and intermediate grade chondrosarcomas signi®cantly higher in ZK-58 and Saos-2 (both may simply be due to low cellularity; however, this P=0.009). In contrast the number of trisomies was reason cannot be used to explain the lack of hTERT signi®cantly higher in MNNG (P=0.0001). gene expression in highly cellular, TA negative osteosarcomas. In addition to hTERT, hTR also appeared to be TP53 status related to TA. With the exception of U2-OS, all Osteosarcoma cell lines OST, SJSA-1 and ZK-58 tumors and cell lines did express hTR with a tendency harbored no point mutations in the coding region of towards higher expression levels in telomerase-positive the TP53 gene by sequencing using the GeneChip TP53 tumor cells. These data support the ®nding that up- Assay (Table 3). regulation of hTR is necessary for TA (Mitchell et al., 1999). Moreover, because hTR expression was found in TA negative osteosarcoma cell lines and primary Discussion tumors, the lack of telomerase activity in these cells cannot be explained by an abrogation of hTR. Approximately 40% of the malignant bone tumors The equal frequency of high grade osteosarcomas included in the study displayed no TA, consistent with and lacking TA, together with the ®nding of ALT with earlier investigations (Aue et al., 1998). When in the osteosarcoma cell lines U2-OS and ZK-58 and in the malignant tumors were grouped according to primary tumors, suggest that TA is not mandatory for their histologic subtype there were dierences in the tumor progression in this entity. frequency of TA: whereas only half of the As expected, in telomere length analysis, the TA osteosarcoma tumor samples were positive for TA, positive cell lines MNNG, HOS, OST and SJSA-1 70% of the chondrosarcomas displayed TA. Remark- exhibited short telomeres of homogeneous length

Oncogene ALT is associated with chromosomal instability C Scheel et al 3840 consistent with most malignant immortalized cells hypothesis is provided by our preliminary M-FISH examined to date (Wright et al., 1996). Semi- data. Furthermore, we are in the process of con®rming quantitative analysis of telomere morphology of OST these results with additional telomerase-negative tumor by T-FISH re¯ected these ®ndings by showing four cell lines exhibiting an ALT phenotype. A function of ¯uorescent spots of similar intensity at each telomeric telomerase in `healing' broken chromosome ends, as position. A similar morphology was also found in the proposed by Melek and Shippen (1996), would provide telomerase positive neuroblastoma cell line Lan-5 and another aspect why ALT cells have features of a normal human lymphocyte culture which were chromosomal instability. included into this study as controls. Furthermore, increasing data suggests that rescue of In contrast, osteosarcoma cell line Saos-2 proved to telomeric function is not only determined by net have low telomerase activity and elongated telomeres lengthening of telomeres. Zhu et al. (1996) proposed typical for the ALT phenotype. It is possible that the a requirement for capping of telomeres by telomerase, low telomerase activity found in Saos-2 is not sucient another function that might be abrogated in ALT cells. to prevent telomere shortening which in turn could give Although ALT appears only in a subset of tumors, rise to ALT. Therefore, we referred Saos-2 to be an particularly of mesenchymal origin, its direct associa- ALT cell line of intermediate type. Perrem et al. (1999) tion with telomeric dysfunction may have important demonstrated that the ALT telomere phenotype was implications for the valuation of chromosomal in- abrogated in cell hybrids between ALT and telomerase- stability in tumor progression. ALT does not seem to positive cells, implying that co-existence of ALT and be an equivalent to telomerase in telomere main- telomerase activity is rather unlikely. However, the tenance; the requirement of telomere stabilization in nature of ALT remains unclear and dierent mechan- tumor progression might be dependent on preceding isms or conditions can be used to describe the same mutations during tumorigenesis. Further studies are phenomenon. needed to analyse if tumor cells that are originally TA T-FISH ¯uorescence intensity corresponds directly positive may become TA negative ALT cells when to telomere length (Lansdorp et al., 1996). ALT cell treated with the emerging group of small molecule lines U2-OS and ZK-58 as well as Saos-2 revealed drugs able to inhibit telomerase activity. intracellular and inter-chromosomal variations in telomere morphology, implying that the heterogeneity of telomere length found by Southern blotting TRF analysis is not simply an intercellular phenomenon due Materials and methods to a mixed population composed of cells with long Materials telomeres and cells with short telomeres. The long and hyper-variable telomeres found in the ALT cell lines Tumor specimens were derived from snap-frozen material of are reminiscent of telomerase-null survivor cells in primary tumor biopsies and surgical resections and stored at yeast which maintain their telomeres through non- 7808C until use. Clinicopathological data comprises diagnosis, sex, age at diagnosis and tumor cell content in the reciprocal recombination (Kass-Eisler and Greider, obtained specimens. The patient's mean age at diagnosis was 2000). Furthermore, the ALT cell lines we analysed 21+13 years for the osteosarcomas, 54+13 years for the presented not only intracellular polymorphism in chondrosarcomas and 32+18 years for the benign lesions. telomere length but also a spectrum of unusual Tumor cell content in relation to the whole tissue area chromosomal loci and con®gurations of telomeric (matrix) evaluated by histologic study ranged from 5 ± 100% repeats with the occurrence of long, dicentric marker in malignant tumors and 5 ± 80% in the benign lesions. chromosomes. Furthermore, osteosarcoma cell lines HOS, MNNG, OST, Telomeres are important for chromosomal stability SJSA-1, Saos-2, U-2 OS and ZK-58, and the neuroblastoma (Counter et al., 1992; Autexier and Greider, 1996). A cell line Lan-5 were included in the study. peak of dicentric chromosomes resulting mostly from telomeric associations occurs in SV40 transformed Protein preparation and telomerase assay ®broblasts prior to crisis when telomere shortening Extractions of cellular proteins of the tumor specimens were reaches a critical point (Ducray et al., 1999). Activated prepared as previously described (Poremba et al., 1998). telomerase homogenizes telomeres in a dynamic Osteosarcoma cell lines were washed with cold Phosphate balance between elongation and shortening towards a Buered Saline (PBS) prior to extraction. The protein stabilized length which in turn promotes chromosomal concentrations were measured by use of the Coomassie stability. Therefore, we suggest that ALT cells do not Protein Assay Reagent (Bio Rad, USA) on an ELISA Reader expand their replicative potential under telomere- and adjusted to about 2 mg/ml. stabilizing conditions. The occurrence of elongated For in vitro detection of TA, a modi®ed version of the and shortened telomeres and signal-free chromosome telomeric repeat ampli®cation protocol (TRAP) assay includ- ing an internal PCR-ampli®cation control from a commer- ends could give rise to chromosomal end-to-end cially available kit (TRAPeze telomerase detection kit, Oncor, associations and breakage-fusion-bridge cycles Gaithersburg, MD, USA) was used, resulting in a 36-bp (McClintock, 1941; Autexier and Greider, 1996; product which was co-ampli®ed with telomerase-elongated Bouer et al., 1996) resulting in an increased number products in a competitive manner (Poremba et al., 1999). of complex non-reciprocal chromosomal rearrange- Each analysis comprised a telomerase positive control cell ments in TA negative cells. Evidence to support this line included in the kit, a heat-inactivated control (telomer-

Oncogene ALT is associated with chromosomal instability C Scheel et al 3841 ase-positive control incubated at 858C for 10 min prior to RNA was ampli®ed using the primers hTR-1 (5'-CCTAACT- reaction) and a negative control (CHAPS lysis buer instead GAGAAGGGCGTAGGC-3') corresponding to GenBank of sample protein). Electrophoresis and semiquantitative positions 849 ± 870 and hTR-2 (5'-CTAGAATGAACGGTG- analysis of chromatogram peaks generated from photode- GAAGGCG-3') corresponding to positions 961 ± 940 (Gen- tector signals were performed as described (Poremba et al., Bank accession number AF047386). The reaction conditions 1998) on an automated laser-¯uorescence sequencer (ALFex- were initial denaturation at 958C for 2 min followed by 35 ± press, Pharmacia, Freiburg, Germany). TRAP data was 40 cycles of denaturation at 958C for 1 s, annealing at 608C analysed in blind-trial fashion. For each tumor sample, the for 5 s and extension at 728C for 6 ± 8 s (depending on TRAP procedure was done twice and levels of telomerase ampli®cation length). Quantitative analysis was performed activity proved to be consistent in all the samples included in using the LightCycler Software employing a real-time this study. To rule out the possibility of telomerase and Taq- ¯uorogenic detection system for a kinetic, rather than end- polymerase inhibiting factors, mixed tissue samples that point approach as on conventional agarose or polyacrylamide contained telomerase-negative and positive extracts were gels. The generation of quantitative data was based on the prepared. Interpretation of proportions distinguished between dierent PCR kinetics of samples with dierent levels of absent TA, low (530% of positive control), intermediate target gene expression. We employed a relative quanti®cation (30 ± 70% of positive control) and high TA (470% of (Poremba et al., 2000) in which the expression levels of the positive control) as previously described (Poremba et al., tumor samples and osteosarcoma cell lines were compared to 1998). the data from the Ewing's tumor telomerase positive cell line VH-64 in a geometric dilution series (1 : 1, 1 : 2, 1 : 4, 1 : 8, 1 : 16, 1 : 32). The graph of the linear regression and Assessment of tissue sample preservation by RT ± PCR of RNA calculation of the regression coecient, r, served to con®rm content accuracy and reproducibility of this approach. For this In order to minimize the probability of false-negative results, approach, the identity and speci®city of the PCR product was with a lack of or low TA due to tissue degradation and con®rmed by melting curve analysis which is part of the necrosis, RNA derived from frozen sections was ampli®ed by LightCycler analysis program. The speci®c melting point of reverse-transcriptase PCR (RT ± PCR) for glyceraldehyde-3- the PCR product was correlated with its molecular weight phosphate dehydrogenase (GAPDH) as an indirect marker of determined by agarose gel electrophoresis. tissue integrity. Brie¯y, a 297-bp fragment of the human Evaluation of hTERT expression in the seven osteosarcoma GAPDH gene was ampli®ed with primers 5'-CACCCATGG- cell lines was performed by using the LightCycler Telo- CAAATTCCATGGC-3' and 5'-GCATTGCTGATGAT- TAGGG-Kit (Roche Diagnostics, Mannheim, Germany) CTTGAGGCT-3', correponding to GenBank positions which comprises a one-step RT ± PCR in which relative 213 ± 234 and 509 ± 487, respectively (GenBank Accession quanti®cation is realized by normalizing hTERT gene Number M33197). For a small subset of tumors, no RNA expression on expression of PBGD (porphobilinogen deami- was available. For those lesions, viability of the TRAP assay nase). For the ampli®cation, 2-ml prediluted aliquots contain- was veri®ed by tumor cell count on hematoxylin-eosin stained ing approximately 200 ng of total RNA were subjected to histologic slides. ®rst strand-synthesis and subsequent ampli®cation in a ®nal volume of 20 ml containing 16ready-to-use Reaction Mix, 0.1 ml reverse transcriptase, 16hTERT Detection Mix or Quantitative real-time RT ± PCR for expression of hTERT and PBGD reaction mix and a ready-to-use primer and hTR hybridization probe mixture speci®c for either hTERT or Total RNA was isolated from fresh-frozen tissue using the PBGD mRNA. Probes consist of two dierent ¯uorescein- TRISOLV system (Biozol, Eching, Germany) in accordance labeled short oligonucleotides that hybridize to an internal with the manufacturer's protocol. RNA was treated with sequence of the ampli®ed fragment during the annealing DNAse (Eurogentec, Seraing, Belgium) and puri®ed by phase and emit a ¯uorescence signal in close proximity repeating the TRISOLV extraction protocol. cDNA was through ¯uorescence resonance energy transfer (FRET). Pre- synthesized from approximately 2 mg of RNA using the diluted RNA standards for establishing a reference curve as FirstStrand Synthesis Kit (AmershamPharmacia, Freiburg, well as a positive control RNA to ensure reliability and Germany) with dT18 primers. Relative concentrations of reproducibility were included in the kit. cDNA samples were evaluated by quantitative RT ± PCR of GAPDH performed on the LightCycler (Roche Diagnostics, DNA extraction Mannheim, Germany) followed by analysis of the gene expression of the telomerase catalytic subunit hTERT and the Genomic DNA was prepared by proteinase K digestion RNA component hTR. For ampli®cation of the cDNA, 5 ml according to the Puregene DNA extraction protocol (Biozym, aliquots of reverse-transcribed cDNA were subjected to PCR Hess. Oldendorf, Germany) followed by treatment with in 20 ml containing a ®nal concentration of 2 ± 3 mM MgCl2, RNase A. 0.5 mM of each primer, and 2 ml of ready-to-use LightCycler DNA Master SYBRGreen I (106, containing TaqDNA Telomere length analysis polymerase, reaction buer, dNTP mix with dUTP instead of dTTP, SYBRGreen I dye, and 10 mM MgCl2). For hot-start Telomeric restriction fragment (TRF) length analysis was PCR, 0.16 ml/sample of TaqStart Antibody (Clontech, carried out using the Telomere Length Assay Kit (Pharmin- Heidelberg, Germany) was added to the ampli®cation gen, San Diego, CA, USA). In brief, 5 mg of genomic DNA mixture prior to the addition of primers and template cDNA. was digested with a HinfI/RsaI restriction enzyme mixture. hTERT mRNA was ampli®ed using the primers hTERT-1 Digested DNA was loaded onto 0.6% agarose gels and run at (5'-CGGAAGAGTGTCTGGAGCAA-3') corresponding to 5 V/cm for approximately 3 h. BstEII and HindIII digested l GenBank positions 1785 ± 1804 and hTERT-2 (5'-CATG- DNA supplied with the kit were used as length standards. GACTACGTCGTGGGAG-3') corresponding to positions Southern transfer blot onto a positively charged nylon 1961 ± 1980 (GenBank Accession number AF018167). hTR membrane (Roche Diagnostics, Mannheim, Germany) was

Oncogene ALT is associated with chromosomal instability C Scheel et al 3842 realized by alkaline vacuum blotting using the VacuGene XL Separate digitized gray level images of DAPI, FITC and blotting unit according to the manufacturer's protocol rhodamine ¯uorescence were taken with a CCD camera (Amersham Pharmacia). Membranes were then hybridized connected to a Leica DMRBE microscope. The image overnight at 558C with 5 ng/ml of an 51-mer biotinylated processing was carried out by use of Applied Image Software. telomere probe (Pierce Chemical Co., USA). Stringency Average green-red ratios were calculated for each chromo- washes were carried out by washing membranes three times some in 5 ± 10 metaphases. for 5 min in 26SSC/0.1% SDS at room temperature. Chromosomal regions with CGH ratio pro®les surpassing Chemiluminescent detection was performed with Streptavi- the 50% CGH thresholds (upper threshold 1.25, lower din-HRP (1 : 300) with solutions and in accordance to threshold 0.75) were de®ned as loci with copy number gains instructions supplied with the Telomere Length Assay kit. or losses. Based on experiments with normal control DNA, Exposure of an X-ray ®lm (X-omat AR, Kodak, Rochester, these thresholds have been proved to eliminate false-positive NY, USA) to the membrane revealed telomere lengths as results. As the tumor specimens and normal DNA were not smears ranging from 1.9 to more than 23.1 kilobases (kb). sex matched, entire X and Y chromosomes were excluded. TRF length analysis was performed on a densitometric scan of the autoradiogram, calculating the mean TRF length for Multiplex-fluorescence in situ hybridization (M-FISH) each sample by integrating the signal intensity above background over the entire TRF distribution as a function Metaphase spreads from the osteosarcoma cell lines MNNG, of TRF length (L=S(ODi6Li)/S(ODi), where ODi and Li Saos-2 and ZK-58 were prepared according to standard are the signal intensity and TRF length respectively at protocols. M-FISH was performed as described (Speicher et position i on the gel image). al., 1996; Eils et al., 1998) with minor modi®cations. Brie¯y, ®ve pools of whole chromosome painting probes (kindly provided by M Ferguson-Smith, Cambridge, UK) were Telomere fluorescence in situ hybridization (T-FISH) ampli®ed and labeled by DOP-PCR using ®ve dierent T-FISH analysis was carried out using the DAKO Telomere ¯uorochromes FITC, Cy3, Cy3.5, Cy5 and Cy5.5, respec- PNA FISH Kit/Cy3 (DAKO, Glostrup, Denmark). Meta- tively). About 100 ng of each chromosome painting probe phase spreads were prepared according to standard proce- was precipitated in the presence of 30 mg Cot-1 DNA, dures and immersed in Tris-buered Saline (TBS) for 5 min resolved in 10 ml hybridization mixture (15% dextrane prior to ®xation in 3.7% formaldehyde in TBS for 2 min, sulfate, 26SSC) and hybridized for 48 h. washes in TBS for 265 min and treatment with proteinase K Microscopic evaluation was performed using a Leica (DAKO, Glostrup, Denmark) for 10 min. After further DMRXA-RF8 microscope (Leica, Wetzlar, Germany) washes in TBS for 265 min, slides were dehydrated with a equipped with a Sensys CCD camera (Photometrics, Tucson, cold ethanol series and air-dried. Ten microliters of Cy3- AZ, USA) with a Kodak KAF 1400 chip. Images for each conjugated telomere speci®c peptide nucleic acid (PNA) ¯uorochrome were aquired separately using highly speci®c probe in hybridization solution containing 70% formamide ®lter sets (Chroma Technology Corp., Brattleboro, VT, USA) was added and DNA was denatured by heat for 3 min at and processed using the Leica MCK software (Leica 808C. After hybridizing in the dark for 30 min, slides were Microsystems Imaging Solutions Ltd., Cambridge, UK). brie¯y immersed in Rinse Solution (supplied with the kit) and then washed at 658C for 5 min in Wash Solution (supplied GeneChip p53 assay with the kit). After rinsing with TBS slides were counterstained with DAPI, and mounted with antifade solution DNA extracted from osteosarcoma cell lines OST, SJSA-1 (VectaShield, Vector laboratories Inc., Burlingame, CA, and U2-OS was sequenced using the GeneChip p53 assay USA). (Aymetrix, Santa Clara, CA, USA) according to the Separate digitized gray level images of DAPI and Cy3 manufacturer's protocol. TP53 exons 2 ± 11 from each cell ¯uorescence were taken with a charge coupled device (CCD) line DNA were ampli®ed as 10 separate amplicons in a camera (Cohu 6X-924) connected to a Leica DMRBE decaplex PCR. Each 100-ml PCR contained PCR Buer II microscope. The image processing was carried out by use (Perkin Elmer), 2.5 mM MgCl2 (Perkin Elmer), 0.2 mM of of Applied Image Software. each dNTP (Sigma), 5 ml of the p53 Primer Set (Aymetrix), 10 U of AmpliTaq GoldTM (Perkin-Elmer), and 250 ng of genomic DNA. PCR was carried out on MJ Research, Inc., Comparative genomic hybridization (CGH) thermocyclers under the following reaction conditions: initial CGH analysis, microscopy and digital image analysis was denaturation at 958C for 10 min, followed by 35 cycles of performed as previously described (Brinkschmidt et al., 1998). 958C for 30 s, 608C for 30 s, and 728C for 45 s, with a ®nal Brie¯y, tumor or cell line DNA was labeled with biotin-16- extension at 728C for 10 min. Forty-®ve-ml volumes of tumor dUTP (Roche Diagnostics, Mannheim, Germany) and or reference amplicons were subjected to fragmentation using reference DNA from a healthy male donor was labeled with 0.25 U of GeneChip Fragmentation Reagent (Aymetrix) at digoxigenin-11-dUTP (Roche Diagnostics, Mannheim, Ger- 258C for 15 min in 0.4 mM EDTA (Gibco ± BRL Life many) in a standard nick reaction. For CGH, 500 ng of the Technologies), 2.5 U of calf intestine alkaline phosphatase tumor DNA, 500 ng of reference DNA and 30 mg of human (Amersham-Pharmacia Biotech), and 0.5 mM Tris.acetate Cot 1 DNA (Gibco) were used for hybridization on target (pH 8.2) (Teknova) before heat inactivation at 958C for metaphase spreads (46,XY) as described elsewhere. Post- 10 min. hybridization washes were carried out to a stringency of 50% The fragmented tumor or reference DNAs (50 ml) were 3' formamide/26SSC at 458C and 0.16SSC at 608C. Biotiny- end labeled with ¯uorescein-N6-dideoxy-ATP. Twenty-®ve U lated and digoxigenated sequences were detected simulta- of terminal transferase (Promega), TdTase Buer (Promega), neously, using avidin-FITC (Roche Diagnostics, 1 : 200) and 10 mM ¯uorescein-N6-ddATP (DuPont/NEN), and the frag- anti-digoxigenin-rhodamine (Roche Diagnostics, 1 : 40). The mented DNA were incubated at 378C for 45 min in a 100-ml slides were counterstained with DAPI and mounted with reaction followed by heat inactivation at 958C for 5 min. The antifade solution (Vectashield, Vector laboratories Inc.) ¯uorescein-labeled DNAs were hybridized to the GeneChip

Oncogene ALT is associated with chromosomal instability C Scheel et al 3843 p53 probe arrays using the GeneChip Fluidics Station 400. Acknowledgments 0.5-ml reaction volumes containing 6X SSPE (0.9 M NaCl, We would like to thank Dagmar Haves, Frauke Schmidt, 0.06 M NaH2PO4,6mM EDTA) (Bio Whitakker), 0.05% Petra Meier and Lydia Grote for excellent technical Triton1 X-100 (Sigma), 1 mg of acetylated BSA (Sigma), assistance. We would also like to thank Andreas Scheel 2nM Control Oligonucleotide F1 (Aymetrix), and the for useful advice. Supported in part by the German labeled DNAs were hybridized at 458C for 30 min. Each Research Foundation ``Deutsche Forschungsgemeinschaft probe array was washed twice with Wash Buer A (36SSPE, DFG'' (grant nos. Po. 529/4-1 and Po. 529/5-1) and the 0.005% Triton1 X-100) at 358C prior to being scanned on a IZKF Muenster (grant nos. G2 and H2). HP GeneArray Scanner (Hewlett-Packard). The captured light signals were analysed using GeneChip Software.

References

Aue G, Muralidhar B, Schwartz HS and Butler MG. (1998). Landers JE, Cassel SL and George DL. (1997). Cancer Res., Ann. Surg. Oncol., 5, 627 ± 634. 57, 3562 ± 3568. Autexier C and Greider CW. (1996). Trends Biochem. Sci., Lansdorp PM, Verwoerd NP, van de Rijke FM, Dragowska 21, 387 ± 391. V, Little MT, Dirks RW, Raap AK and Tanke HJ. (1996). Avilion AA, Piatyszek MA, Gupta J, Shay JW, Bacchetti S Hum. Mol. Genet., 5, 685 ± 691. and Greider CW. (1996). Cancer Res., 56, 645 ± 650. McClintock B. (1941). Genetics, 26, 234 ± 282. Biessmann H and Mason JM. (1997). Chromosoma, 106, 63 ± Melek M and Shippen DE. (1996). Bioessays, 18, 301 ± 308. 69. Meyerson M, Counter CM, Eaton EN, Ellisen LW, Steiner Bouer SD, Morgan WF, Pandita TK and Slijepcevic P. P, Caddle SD, Ziaugra L, Beijersbergen RL, Davido MJ, (1996). Mutat. Res., 366, 129 ± 135. Liu Q, Bacchetti S, Haber DA and Weinberg RA. (1997). Brinkschmidt C, Poremba C, Christiansen H, Simon R, Cell, 90, 785 ± 795. Schafer KL, Terpe HJ, Lampert F, Boecker W and Mitchell JR, Wood E and Collins K. (1999). Nature, 402, Dockhorn-Dworniczak B. (1998). Br. J. Cancer, 77, 551 ± 555. 2223 ± 2229. Olovnikov AM. (1971). Dokl Akad Nauk., 201, 1496 ± 1499. Broccoli D, Young JW and de LT. (1995). Proc. Natl. Acad. PerremK,BryanTM,EnglezouA,HacklT,MoyELand Sci. USA, 92, 9082 ± 9086. Reddel RR. (1999). Oncogene, 18, 3383 ± 3390. Bryan TM, Englezou A, Dalla-Pozza L, Dunham MA and Poremba C, Bocker W, Willenbring H, Schafer KL, Reddel RR. (1997). Nat. Med., 3, 1271 ± 1274. Otterbach F, Burger H, Diallo R and Dockhorn DB. Bryan TM, Englezou A, Gupta J, Bacchetti S and Reddel (1998). Int. J. Oncol., 12, 641 ± 648. RR. (1995). EMBO J., 14, 4240 ± 4248. PorembaC,ScheelC,HeroB,ChristiansenH,SchaeferKL, Bryce LA, Morrison N, Hoare SF, Muir S and Keith WN. Nakayama J, Berthold F, Juergens H, Boecker W and (2000). Neoplasia, 2, 197 ± 201. Dockhorn-Dworniczak B. (2000). J. Clin. Oncol., 18, Chen PL, Chen YM, Bookstein R and Lee WH. (1990). 2582 ± 2592. Science, 250, 1576 ± 1580. Poremba C, Shroyer KR, Frost M, Diallo R, Fogt F, SchaÈ fer Counter CM, Avilion AA, LeFeuvre CE, Stewart NG, K, BuÈ rger H, Shroyer AL, Dockhorn-Dworniczak B and Greider CW, Harley CB and Bacchetti S. (1992). EMBO Boecker W. (1999). J. Clin. Oncol., 17, 2020 ± 2026. J., 11, 1921 ± 1929. Radig K, Schneider-Stock R, Haeckel C, Neumann W and Counter CM, Gupta J, Harley CB, Leber B and Bacchetti S. Roessner A. (1998). Hum. Pathol., 29, 1310 ± 1316. (1995). Blood, 85, 2315 ± 2320. Ramakrishnan S, Eppenberger U, Mueller H, Shinkai Y and Counter CM, Hahn WC, Wei W, Caddle SD, Beijersbergen Narayanan R. (1998). Cancer Res., 58, 622 ± 625. RL, Lansdorp PM, Sedivy JM and Weinberg RA. (1998). Romano JW, Ehrhart JC, Duthu A, Kim CM, Appella E and Proc. Natl. Acad. Sci. USA, 95, 14723 ± 14728. May P. (1989). Oncogene, 4, 1483 ± 1488. Ducray C, Pommier JP, Martins L, Boussin FD and Sabatier Shay JW. (1999). J. Natl. Cancer Inst., 91, 4±6. L. (1999). Oncogene, 18, 4211 ± 4223. Shay JW and Bacchetti S. (1997). Eur. J. Cancer, 33, 787 ± Eils R, Uhrig S, Saracoglu K, Satzler K, Bolzer A, Petersen I, 791. Chassery J, Ganser M and Speicher MR. (1998). Speicher MR, Gwyn BS and Ward DC. (1996). Nat. Genet., Cytogenet. Cell. Genet., 82, 160 ± 171. 12, 368 ± 375. Greider CW and Blackburn EH. (1987). Cell, 51, 887 ± 898. Sumida T, Hamakawa H, Sogawa K, Sugita A, Tanioka H Harle BC and Boukamp P. (1996). Proc. Natl. Acad. Sci. and Ueda N. (1999). Int. J. Cancer, 80, 1±4. USA, 93, 6476 ± 6481. Takakura M, Kyo S, Kanaya T, Tanaka M and Inoue M. Hiyama E, Hiyama K, Tatsumoto N, Kodama T, Shay JW (1998). Cancer Res., 58, 1558 ± 1561. and Yokoyama T. (1996). Int. J. Oncol., 9, 453 ± 458. Takubo K, Nakamura K, Izumiyama N, Mafune K, Tanaka Hiyama K, Hirai Y, Kyoizumi S, Akiyama M, Hiyama E, Y, Miyashita M, Sasajima K, Kato M and Oshimura M. Piatyszek MA, Shay JW, Ishioka S and Yamakido M. (1997). J. Surg. Oncol., 66, 88 ± 92. (1995). J. Immunol., 155, 3711 ± 3715. Tanaka H, Shimizu M, Horikawa I, Kugoh H, Yokota J, Ito H, Kyo S, Kanaya T, Takakura M, Inoue M and Namiki Barrett JC and Oshimura M. (1998). Genes Chromosomes M. (1998). Clin. Cancer Res., 4, 1603 ± 1608. Cancer, 23, 123 ± 133. Kass-Eisler A and Greider CW. (2000). Trends Biochem. Sci., Watson JD. (1972). Nat. New Biol., 239, 197 ± 201. 25, 200 ± 204. Wright WE, Brasiskyte D, Piatyszek MA and Shay JW. Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, (1996). EMBO J., 15, 1734 ± 1741. Ho PL, Coviello GM, Wright WE, Weinrich SL and Shay JW. (1994). Science, 266, 2011 ± 2015.

Oncogene ALT is associated with chromosomal instability C Scheel et al 3844 Wright WE, Pereira SO, Shay JW, et al. (1989). Mol. Cell. Zhu X, Kumar R, Mandal M, Sharma N, Sharma HW, Biol., 9, 3088 ± 3092. Dhingra U, Sokoloski JA, Hsiao R and Narayanan R. Wynford-Thomas D. (1999). J. Pathol., 187, 100 ± 111. (1996). Proc. Natl. Acad. Sci. USA, 93, 6091 ± 6095.

Oncogene