Multiple Mechanisms of Telomere Maintenance Exist in Liposarcomas Free

Telomeres are found at the ends of DNA molecules and are required for the maintenance of the linear chromosomes found in eukaryotes. These specialized structures protect against chromosome end-to-end fusion events and prevent the ends of DNA molecules from being recognized by DNA damage-sensing mechanisms (1). Telomeric sequences are lost each time a cell replicates its DNA. Thus, telomere length in mitotic tissues, such as those of the hematopoietic system or skin fibroblasts, is linked to the age of the individual (2). The telomere attrition that accompanies somatic cell division eventually compromises the telomeric end-protective complex and the resulting telomere dysfunction activates DNA damage checkpoints (3, 4) leading to cellular responses such as apoptosis (5) or senescence (6). This telomere length–dependent block to continual proliferation acts to limit the likelihood that an aberrantly growing cell will develop into a tumor. However, if cellular genome surveillance pathways are compromised (e.g., by p53 mutation), continued cell division with dysfunctional telomeres results in massive genome instability (7). Data from mouse models of tumorigenesis suggest that a transient period of genome instability facilitates carcinogenesis by promoting mutagenecity (7, 8). Similarly, telomere-driven genome instability may promote the development of human breast cancer (9). Continual, unlimited growth of cells can be achieved, however, by the activation of a telomere maintenance mechanism (10, 11). The associated telomeric stabilization allows evasion of cell cycle checkpoints thereby facilitating the unchecked proliferation necessary for tumor formation. Although it is clear that activation of a telomere maintenance mechanism facilitates the cellular immortalization associated with tumorigenesis, recent evidence suggests that telomere maintenance is not absolutely required for initial tumor formation in all settings (12).

Stabilization of the telomeric DNA array is usually accomplished by the action of a large, multi-subunit ribonucleoprotein complex, known as telomerase, which adds telomeric repeats to a preexisting telomere terminus (13). The human telomerase holoenzyme is composed of two essential subunits, a telomere-specific reverse transcriptase (hTERT) and an RNA template (hTR), and functions as an interdependent dimer or multimer (14). Whereas the majority of human epithelial tumors use telomerase for the constitutive maintenance of the telomeric repeat array, a subset of tumors employ a telomerase-independent mechanism called alternative lengthening of telomeres (ALT; refs. 15, 16). The specific genetic determinants required for ALT are unknown; however, there are several key differences between ALT-positive and telomerase-positive cells. For example, in contrast to the fairly homogenous telomeres found in telomerase-positive cells, the telomeres in most ALT-positive cells are quite heterogenous, ranging from 4 to >25 kb in size (15). Another feature of ALT cells is that they contain extrachromosomal DNA circles composed of telomeric repeats which are readily detected by both pulsed-field gel electrophoresis and electron microscopy (17, 18). ALT-positive cells also differ from telomerase-positive cells in that ALT-type cells show a cell cycle-dependent colocalization of telomeric DNA, the telomeric-binding proteins TRF1 and TRF2, and the promyelocytic leukemia (PML) nuclear body in a complex called the ALT-associated PML nuclear body (APB; refs. 19–21). Factors involved in recombination and the cellular DNA damage response are also present at APBs (19, 21, 22). The PML nuclear body is a large multiprotein structure that is involved in a variety of cellular processes, including apoptosis, differentiation, and growth control (23, 24). Although the specific function(s) of APBs are not known, the presence of these structures is tightly linked with ALT and can serve as a diagnostic marker for cells that use this mechanism for telomere maintenance.

Recently, it was shown that a significant proportion of osteosarcomas (35%; ref. 25) and glioblastoma multiforme (19%; ref. 26) use ALT for telomere maintenance. In fact, the level of ALT activation observed in these tumors was comparable, if not slightly higher than what was seen for telomerase (18% and 29%, respectively; refs. 25, 26). Additional studies of multiple subtypes of sarcoma based primarily on presence of APBs are consistent with an increased incidence of ALT activation compared with that observed in carcinomas (27, 28). It has been suggested from investigations in telomerase-deficient mice that ALT-positive tumors are less aggressive than telomerase-positive tumors (29). Consistent with this, glioblastoma multiforme patients with ALT-positive tumors were found to have a better prognosis than those whose tumors used telomerase for telomere maintenance (26). However, this is not always the case, as a similar association was not observed for osteosarcomas (25) or for soft tissue sarcomas (STS; ref. 28). Interpretation of the role of telomere maintenance mechanism in STS has been hampered by having a sample set composed of tumors of different biology and from different tissues. Furthermore, the effect of telomere maintenance mechanism on disease progression likely varies with tumor type. It is notable that in sarcomas associated with diagnostic translocations, the frequency of ALT is reduced relative to sarcomas not associated with a translocation (27). Although it is evident that the mutations driving tumorigenesis in these two categories of liposarcoma are different, the molecular basis of the differential activation of ALT in these tumor types is unknown.

In the current study, we present analysis of telomere maintenance mechanisms in a panel of the most common STS, liposarcoma (30). As in previous studies, tumor samples were analyzed with respect to telomerase activity and telomere length. To improve the accuracy of categorization, we also assayed cells for the presence of ALT-specific subcellular structures, APBs. Our data reveal a similar incidence of ALT and telomerase-positivity in liposarcomas, whereas evidence of both mechanisms being active in the same tumor was obtained from one sample. In addition, a large proportion of these tumors did not have characteristics of telomerase or ALT, suggesting that such tumors either employ a novel constitutively active telomere maintenance mechanism or lack such a mechanism. These findings have important implications for understanding the role of telomere maintenance in the development of liposarcoma as well as designing strategies to treat these tumors.

Liposarcoma samples. The Fox Chase Cancer Center has an Institutional Review Board–approved protocol to bank blood samples and excess tissue from patients undergoing surgical resections. Tumor samples are made available with relevant clinical history with all patient identifiers removed. The Fox Chase Cancer Center Tumor Bank Facility conducted a search for liposarcoma samples. Each sample was reviewed and graded using standard criteria (31). After reviewing H&E-stained paraffin sections, 34 cases of liposarcomas were selected. The respective frozen blocks embedded in optimal cutting temperature compound (Sakura Finetek, Torrance, CA) were sectioned to confirm the presence of liposarcoma and used for immunofluorescent analysis. Frozen tissue for each tumor specimen was also obtained from the Fox Chase Cancer Center Tumor Bank Facility and used to prepare cellular extracts, genomic DNA, and total RNA.

Telomeric repeat amplification protocol assay. The telomeric repeat amplification protocol (TRAP) assay was done essentially as previously described (32). Briefly, cellular extract was prepared by homogenizing 150 mg of tumor tissue in CHAPS extraction buffer. Following incubation on ice for 30 minutes, the suspension was centrifuged at 14,000 rpm for 15 minutes at 4°C and the supernatant collected for use in the TRAP assay. The pellet was washed twice with 1× PBS and retained for subsequent isolation of genomic DNA for Southern analysis of telomere arrays (see below). Total protein concentrations in the extracts were determined using the Bradford assay (Bio-Rad, Richmond, CA) and for each sample from 0.1 to 1 μg of total protein was assayed for telomerase activity. To control for nonspecific inhibitors present in the extract, all tumor samples were also mixed with telomerase-positive HCT116 or HeLa cells before preparing extract. To ensure that activity was dependent upon telomerase, each reaction was carried out in the presence or absence of RNase, which destroys the RNA component of the holoenzyme. Reaction products were resolved on 15% acrylamide gels and visualized following autoradiography.

Reverse transcriptase-PCR. Approximately 30 mg of tumor tissue were placed in RNAlater (Ambion, Austin, TX) solution prechilled at −80°C and left at −20°C for 18 hours before RNA extraction. Total RNA was extracted using the RNAqueous-4PCR kit (Ambion). Total RNA (0.6 μg) was subjected to reverse transcription and subsequent PCR using the One-step RT-PCR kit (Qiagen, Chatsworth, CA). A reverse transcriptase-PCR (RT-PCR) reaction was carried out to detect a 983-bp product of the glyceraldehyde-3-phosphate dehydrogenase mRNA (33) as a control for the quality of the RNA template used. The hTERT mRNA molecules were amplified as described (33) using primers located at nucleotides 2026 to 2046 and nucleotides 2482 to 2449 for the forward and reverse primers, respectively, that flank a region encompassing the alternative splice sites, thereby resulting in products ranging from 239-bp, representing the smallest alternatively spliced inactive variant, to 457-bp, representing the full-length transcript associated with enzymatic activity. PCR products were resolved on agarose gels and visualized by staining with ethidium bromide (glyceraldehyde-3-phosphate dehydrogenase) or SYBR Green (hTERT).

Southern blotting. Telomere restriction fragment length analysis was done as previously described (32, 34). Genomic DNA was isolated from cellular pellets remaining after CHAPS extraction of telomerase (see TRAP assay) by SDS lysis and phenol extraction. DNA was digested with HinfI and RsaI, resolved on 0.6% agarose gels and transferred to Nytran filters (Amersham, Arlington Heights, IL). Telomeres were visualized following hybridization with an oligonucleotide complimentary to the telomeric repeats (TTAGGG)₄, labeled with T4 polynucleotide kinase and ³²P-γATP. Mean telomere length was determined following densitometric analysis using Telometric software (35).

Indirect immunofluorescence. Liposarcomas were analyzed by indirect immunofluorescence of cryostat sections. In parallel, sections were prepared and stained with H&E and for pathologic analysis (see above). Briefly, samples were subjected to pre-extraction in buffer containing 50 mmol/L NaCl/20 mmol/L HEPES (pH 7.9)/3 mmol/L MgCl₂/0.5% Triton X-100/300 mmol/L sucrose for 5 minutes at room temperature to remove soluble proteins. The samples were washed twice with 1× PBS before being fixed in 3.7% formaldehyde and processed using standard conditions. The telomeric binding protein TRF2 was detected using a mouse monoclonal antibody raised against the full-length human TRF2 protein (IMG-124; Imgenex, San Diego, CA) and the PML nuclear body detected using a goat polyclonal antibody against the NH₂ terminus of the PML protein (N-19; Santa Cruz Biotechnology, Santa Cruz, CA). Primary antibodies were detected with TRITC-conjugated donkey anti-mouse IgG and FITC-conjugated donkey anti-goat IgG (Jackson ImmunoResearch, West Grove, PA). The secondary antibodies did not cross-react. Nuclei were stained with 0.2 μg/mL 4′,6-diamidino-2-phenylindole. Occasional weak 4′,6-diamidino-2-phenylindole staining is an artifact of sectioning. A series of 0.5-μm sections were collected for four to six fields of each tumor using an inverted Nikon TE2000 microscope, equipped for epifluorescence. Image acquisition from a Cascade 650 (Photometrics, Tucson, AZ) monochrome camera was controlled by a PC running the program MetaVue (Universal Imaging/Molecular Devices, Downingtown, PA). An automated Ludl MAC2000 x-y stage and z-axis motor were also controlled using the MetaVue software.

Image analysis. Three-dimensional reconstruction and quantitation of APBs in individual nuclei within each field were carried out using the MetaMorph 6.2r1 image analysis software (Universal Imaging/Molecular Devices). Briefly, a series of z-sections for each channel was reassembled using the “maximum” type option within the “three dimensional reconstruction” function. Nuclei were identified in the 4′,6-diamidino-2-phenylindole channel using the “threshold for light objects” and “create regions from objects” functions. Where necessary, adjacent nuclei were separated into independent regions using the “cut-drawing” tool. Regions were transposed onto the reassembled image for each digital channel, after background was removed using the “flatten background” function and positive signals were identified by manual thresholding. For each nucleus, the percent of the total signal of TRF2 that colocalized with PML was ascertained using the “measure colocalization” function.

Tumor sample characteristics. Thirty-four liposarcoma samples collected between 1992 and 2003 were available for analysis from the Fox Chase Cancer Center Tumor Bank (Table 1). Tumor samples were obtained from males and females with almost equal frequency, and the median age of the patients was 64. The most common site of tumor was the retroperitoneum. Our sample contained a predominance of grade 1 tumors and no grade 3 tumors. Fourteen samples were primary tumors, with the remainder (n = 20) being recurrent tumors bringing the total number of tumors to 34 from 28 cases. We obtained recurrent tumors from two primary tumors (Table 2, tumors 25 and 27). Additionally, we obtained multiple recurrent samples from two patients, although the matched primary tumors were not available (Table 2, tumors 26 and 28). Recurring tumors were found to have a higher grade than the primary tumors, with the exception of tumor 25. This tumor on review was grade 2 but may well have had grade 1 areas that were not sampled. The recurrent tumor was grade 1, either representing sampling error or recurrence of only the low-grade portion. The latter explanation is supported by the 5-year interval between the primary and recurrent tumor.

Analysis of telomerase activity. We have extensively characterized this panel of 34 liposarcomas to determine their active telomere maintenance mechanism(s) (Table 2). Tumors were characterized for presence of telomerase activity using the TRAP assay (ref. 32; Fig. 1A). To confirm that the absence of telomerase activity was not due to a technical artifact or the presence of an inhibitor in the tumor we mixed tissue from the negative tumors with telomerase-positive cells (HCT116 or HeLa) before preparing the extract. In these mixing experiments, the presence of telomerase-positive cells resulted in robust activity (Fig. 1A; data not shown), indicating that the absence of telomerase activity in the tumor sample was not due to technical difficulties or the presence of a diffusible telomerase inhibitor. Using these criteria, 28 of the 34 liposarcomas were informative for telomerase activity, with six tumors scoring positive.

As an alternative assay for telomerase activity, we extracted total RNA from this panel of liposarcomas and carried out RT-PCR to detect mRNA encoding the telomerase catalytic subunit, hTERT (33). In contrast to other components of telomerase, hTERT is directly correlated with enzymatic activity (36, 37). The RT-PCR was done using primers that amplify the mRNA encoding the full-length, enzymatically active hTERT (457 bp), as well as smaller alternatively spliced forms of the message that are not able to maintain telomeric arrays, one of which has been shown to act as a dominant-negative allele. These primers thus span two introns and rule out contaminating DNA as a template source. Amplification of a 983-bp fragment of the mRNA encoding the housekeeping protein glyceraldehyde-3-phosphate dehydrogenase was used to control for the quality of the template RNA (Fig. 1B). RNA suitable for analysis was obtained from 27 of the 34 tumors. The 457-bp hTERT message that encodes enzymatically active protein was readily detectable in the six abovementioned tumors that were positive for telomerase activity (Fig. 1B; data not shown). In addition, the 457-bp product representing message encoding enzymatically active hTERT was detected in three additional tumors that had either been uninformative for or were scored as having undetectable levels of telomerase activity as assessed by the TRAP assay. Importantly, none of these tumors were positive for the dominant-negative splice variant of hTERT. Results from the two assays yielded informative data for 33 of the 34 liposarcoma samples tested here. Therefore, a total of 9 of 33 liposarcomas, representing 27% of the tested tumors, were scored as positive for telomerase using at least one of these two assays.

Telomere length analysis. To further characterize the telomere maintenance mechanism(s) active in each liposarcoma, we carried out Southern analysis of telomere restriction fragments (Fig. 2; data not shown). Sufficient DNA for analysis was obtained from 33 of the 34 tumors (Table 2). Of the nine telomerase-positive tumors scored by enzyme activity or positive RT-PCR, five had telomere arrays averaging ∼7 kb in length, with a range of average length from 6 to 9 kb. Two telomerase-positive tumors had telomere arrays averaging ∼12 kb in length. These data are consistent with the distribution of telomere array lengths seen in a variety of other telomerase-positive human tumors. “ALT-type” telomeres, which are extremely long and heterogeneous ranging in length from <4 to >25 kb (15), were present in two of the telomerase-positive liposarcomas (tumors 9 and 23), an observation potentially indicative of both ALT and telomerase being active simultaneously in a single tumor, highlighting the need for multiple assays to accurately determine the telomere maintenance mechanism(s) that are active.

Telomeres characteristic of ALT (mean telomere length of >15 kb) were present in 6 of the 25 telomerase-negative tumors (Fig. 2; Table 1). Thus, ∼18% (6 of 33 informative samples) of liposarcomas analyzed here would be characterized as solely using ALT for telomere maintenance, based upon analysis of telomere length and telomerase activity. In the remaining 19 telomerase-negative tumors, the telomeres fell within the size range more commonly associated with telomerase-positive samples (<15 kb). Based solely upon the criteria of telomere length and telomerase activity, it is not clear what telomere maintenance mechanism, if any, is used by these liposarcomas.

Analysis of alternative lengthening of telomere–associated promyelocytic leukemia nuclear bodies. To better characterize the tumors that had evidence of both ALT and telomerase activity and those with characteristics of neither pathway, we carried out indirect immunofluorescence analysis to detect the association of telomeric components with the PML nuclear body (APBs; Fig. 3A), structures characteristic of cells that use ALT for telomere maintenance (19). Images were acquired from three to seven independent fields as 0.5-μm sections, which were subsequently reassembled using the three-dimensional reconstruction function in the MetaMorph image analysis software 6.2r1 (Universal Imaging/Molecular Devices). The images were scored to establish if telomeric components, as determined by detection of the constitutive telomeric binding protein TRF2, colocalize with the PML nuclear body. In each nucleus, the percent of total TRF2 signal that colocalized with the total PML signal was quantitated using MetaMorph image analysis software 6.2r1.

To initially validate this approach, we characterized APB frequency in two ALT-positive cell lines, HIO107 and U2OS, as well as in telomerase-positive HeLa cells. A graph of the distribution of the amount of colocalization of TRF2 with PML indicates that ALT-positive and telomerase-positive populations are readily distinguishable (Fig. 3; data not shown). Importantly, in telomerase-positive HeLa cells, >85% of the population has <5% colocalization of the total TRF2 and PML signals. In contrast, in ALT-positive cell lines, <60% of the cells showed colocalization of <5%. These criteria were used to score the tumors as either APB-positive (<65% of cells in the 0-5% colocalization bin) or APB-negative (>80% of cells in the 0-5% colocalization bin). The criteria used here are more stringent than those used in other studies in an effort to most accurately categorize this set of tumors with respect to telomere maintenance mechanism. As a consequence of this, four tumors had an intermediate phenotype with respect to APB status and were unable to be definitively classified.

Using the definitions described above, we find that all six of the telomerase-negative liposarcomas with telomere arrays in excess of 15 kb score positive for APBs. Similarly, we were able to obtain data from three of the telomerase-positive tumors with telomere arrays 12 kb or shorter and in all cases the tumors were scored as APB-negative. These results indicate 100% concordance of the APB-positivity with ALT-positivity and 0% concordance with telomerase positivity. Because we were able to validate this scoring approach with 100% accuracy, we proceeded to clarify which telomere maintenance mechanism (if any) was activated in the remaining tumors that had evidence of both or of neither telomere maintenance pathway being active. Staining of sufficient quality to permit subsequent analysis was obtained from 24 of the original 34 tumors.

In the two telomerase-positive tumors that had telomere arrays averaging >15 kb in length (Table 2, tumors 9 and 23), previously interpreted as both telomere maintenance pathways being active in a single tumor (25, 26), tumor 23 had no evidence of APBs. In the second telomerase-positive tumor, tumor 9, the frequency with which TRF2 and PML colocalized (Table 2, APB{B0-5} = 74%) was intermediate to the extent of colocalization observed in ALT-positive cells (APB{B0-5}< 65%) and telomerase-expressing cells (APB{B0-5}> 80%). Tumor 9 may therefore provide evidence of both ALT and telomerase being active in a single tumor, although this seems a fairly rare occurrence in liposarcomas (1 of 34, 3%). Furthermore, the relative weakness of the APB phenotype in such double-positive tumors may indicate that a low percentage of cells in this tumor use ALT.

In the final category of tumors with no detectable telomerase activity and telomere arrays of <15 kb in length, data from immunofluorescent analysis was obtained from 12 samples. This analysis classified two of these tumors as APB-positive, tumors 3 and 6 with average lengths of 10 and 11.7 kb, respectively, suggesting that these tumors are using ALT for telomere maintenance. Tumors 8 and 11 had a frequency of TRF2 colocalization that was greater than that observed in telomerase-positive samples but that did not reach the cutoff to definitively be classified as ALT and which we, therefore, were unable to classify. The final eight tumors that we were able to analyze here had no evidence of APBs, contained telomeres <12 kb in average length and were telomerase-negative. However, one of these tumors, tumor 14, was not informative for telomerase activity. Thus, in seven tumors, there is no evidence that either of the known telomere maintenance mechanisms, ALT or telomerase, is active.

Activation of a telomere maintenance pathway is tightly linked to cellular immortality and thus to tumorigenesis (38). Therefore, defining the pathways regulating mechanisms of telomere maintenance is critical to the successful management of cancer. The vast majority of carcinomas maintain telomeric DNA arrays by up-regulating the activity of telomerase (16, 39). Indeed, it has been shown that telomerase is essential for unlimited cellular proliferation in vivo (10, 11) and that telomerase inhibition can limit the growth of tumor cells in vitro (40, 41). However, recent studies have suggested that in specific categories of malignancy, such as osteosarcomas (25) and glioblastoma multiforme (26), the frequency with which telomeric DNA is maintained by the ALT pathway is similar to the frequency of telomerase utilization. Additional studies have suggested that this finding may be true for other STS (27, 28), although analysis of a sample set comprised of a single category of STS has not been carried out.

Treatment of soft tissue sarcomas has been hampered by agents with minimal activity as well as significant toxicity, with 50% of patients diagnosed with sarcomas ultimately succumbing to their disease (42). We focused our analysis of telomere mechanisms on liposarcoma because it is the most common STS in adults (30). Selecting patients that would derive benefit from therapy would save patients from enduring toxicity and identify a group that would not only benefit from response to therapy but potentially have prolonged survival. A better understanding of the mechanisms of telomere maintenance would allow rational testing of agents targeting telomerase in liposarcomas and/or provide the impetus for the development of agents specific for tumors that use ALT. Our analysis, unlike prior studies, was done using all available assays of telomere maintenance mechanism. We find that telomerase and ALT are activated at comparable frequencies in this disease (27% and 24%, respectively). Furthermore, we find that a significant proportion of liposarcomas do not have characteristics consistent with the activation of either of the two defined telomere maintenance mechanisms (49%). These findings, together with those of earlier studies, are directly relevant to the effectiveness of treatment strategies that may be employed to limit telomere stabilization and, in so doing, to inhibit tumor growth.

In addition to its role in maintaining chromosome ends, telomerase activation has recently been implicated in providing growth-promoting properties to tumor cells (43–45). This activity of telomerase may be the underlying molecular basis for observations in mouse models suggesting that telomerase-positive tumors are more aggressive and more capable of metastasis than are ALT-positive tumors (29). The association between survival and telomere maintenance mechanism is not so straightforward in human tumors. In the case of glioblastoma multiforme, ALT-positivity is associated with improved survival (26). However, a similar association is not always observed in osteosarcomas (25, 28). This may be due to the differences in both the average age of occurrence and the respective treatments that are available for these malignancies. Indeed, ALT-positive tumors are associated with younger patients for glioblastoma multiforme (26). Previous studies of STS have not permitted these types of analyses because the sample set was composed of a variety of tumor subtypes. In our investigation of a single subtype of STS, liposarcoma, we found that there was no correlation between telomere maintenance mechanism and age. Analysis of a correlation between survival and telomere maintenance mechanism in our sample set could not be carried out in a statistically significant manner due to the small sample size and the older average age of diagnosis for liposarcoma with the increased incidence of age-related chronic illness. The prognostic value of telomere maintenance mechanism may be specific to each type of cancer, possibly reflecting tissue-specific differences in the pathways regulating activation of telomere maintenance mechanisms.

Telomerase activation was not correlated with tumor grade, as roughly equal numbers of telomerase-expressing tumors were grade 1 as grade 2. Similarly, based upon the frequency of ALT activation in grade 1 tumors, this telomere maintenance mechanism was observed in the grade 2 tumors within the expected frequency (1-5 ALT-positive grade 2 tumors; confidence interval, 90%). However, relatively few grade 2 tumors (n = 8) and no grade 3 tumors were analyzed here. Thus, study of additional intermediate and high-grade tumors would more stringently establish the relationship, if any, between telomere maintenance mechanism and tumor grade. Furthermore, such a study would increase the ability to establish if a correlation between telomere maintenance mechanism and survival exists. Such a study may require multi-institution collaboration given the rarity of liposarcoma.

The sample set analyzed here contained four myxoid liposarcomas based on histologic criteria; however, we did not confirm the presence of the diagnostic t(12;16) translocation (46). Two of these tumors used ALT (tumors 1 and 7), one used telomerase (tumor 20) and one tumor did not have characteristics consistent with either telomere maintenance mechanism (tumor 14). In contrast, based upon telomere length and APB positivity, Montgomery et al. (27) found that 78% of sarcomas without diagnostic chromosomal translocation used ALT, whereas 100% of sarcomas with characteristic translocations had activated telomerase. Thus, we did not observe the strong inverse correlation between the frequency of ALT activation and presence of a chromosomal translocation, as expected for myxoid liposarcoma. This may reflect differences in sample composition of the two studies, as the study by Montgomery et al. (27) included two myxoid liposarcomas and no liposarcomas.

It is not clear why in liposarcomas the frequency of tumors that activate ALT is comparable with those that activate telomerase. Identifying the molecular basis of this altered regulation of telomere maintenance mechanisms would be of interest not only for increasing our understanding of telomere metabolism but also for exploring the requirement for activation of telomere maintenance during the development of sarcomas. In addition, these data highlight the fact that a significant fraction of these tumors would be refractory to treatment with anticancer agents that target telomerase. To date, no strategies have been developed aimed at treatment of tumors that use ALT. Therefore, efficient treatment of liposarcoma, as well as other tumor types that use ALT at a relatively high frequency, would benefit from the development of chemotherapeutics that directly target this mechanism and/or the telomere itself.

We found that a number of tumors (n = 7) possessed no apparent mechanism for telomere maintenance, despite being informative for all assays. Thus, these tumors were negative for telomerase, had non-ALT telomeres array lengths ranging in size from 7 to 11 kb, and had APB scores that placed them as non-ALT (B0-5 > 80). Such a finding might be explained in two fashions. Either these tumors truly lack a constitutively active telomere maintenance mechanism, or they employ a mechanism that has not yet been identified. Our analysis of recurrent tumors supports the former possibility. Two cases of telomerase-negative, APB-negative tumors that subsequently recurred were informative for all three phenotypes assayed (telomerase activity, telomere length, and APB class) in this analysis. In one case, tumor 26, telomerase was eventually activated following the second recurrence of the disease, whereas in the second case, tumor 28, the recurrent tumor remained telomerase-negative and had an intermediate APB phenotype. In both instances, average telomere length was shorter in the recurrent tumor than in the earlier occurrence supporting the notion that a mechanism to maintain telomeric arrays was absent, at least transiently, during growth of the recurrent mass. However, it is also important to note that one tumor that was initially telomerase-positive also recurred twice and each time telomeric arrays had an average shorter length despite continued presence of telomerase activity. One possibility is that there was insufficient telomerase activity in this tumor to completely maintain telomeric DNA. Alternatively, telomerase might not have remained active throughout the growth of the recurrent tumors. Because the types of analyses carried out here rely upon single time points, it is not possible to differentiate between these possibilities. Studies analyzing telomere dynamics during in vitro culture of liposarcomas may provide a means of differentiating between the constitutive activity of telomere maintenance mechanisms at levels insufficient to maintain telomeric DNA arrays and mechanisms that alternate between active and inactive states. Such studies are currently not feasible due to the dearth of liposarcoma cell lines. Development of these cell lines would provide valuable reagents for future studies.

There is also a precedent for telomere maintenance mechanisms that defy conventional categorization as telomerase or ALT. A single telomerase-negative immortal cell line (AG11395) has been described in which the terminal DNA array is composed of telomeric repeats interspersed with SV40 sequences (47, 48), indicating that terminal DNA tracts may be composed of repeating units more complex than those found in wild-type telomeres (i.e., as long as they contain a sufficient amount of TTAGGG sequences to recruit capping proteins). This cell line is also unique in that it does not show colocalization of telomeric components with the PML nuclear body, although it does contain multiprotein complexes wherein recombination factors, telomeric factors, and many other APB components are present. Analysis of liposarcoma tumors for such multiprotein structures may be warranted in those samples that do not fit the current classification scheme for telomere maintenance mechanism.

The data presented here raise a number of issues that should be addressed. First, the requirement for telomere maintenance during the development of liposarcomas, and possibly other STS, seems less stringent than for carcinomas where telomerase activation is observed in >80% of tumors (39). This clearly has important implications not only for understanding the mechanisms contributing to STS but also for the development of effective treatment strategies for STS. Furthermore, activation of recombination-based telomere maintenance following inhibition of telomerase has been reported for one colorectal tumor-derived cell line (49). It is not yet known if activation of ALT in response to telomerase inhibition will represent a significant clinical problem in the treatment of STS. However, the prevalence of ALT-positive tumors suggests that this mechanism may be readily activated in STS, making it critical that this issue be addressed, as therapies based upon telomerase inhibition are soon to be in clinical trials. Finally, in this study, we document several incidences of recurrent disease, with associated telomere attrition and absence of a known telomere maintenance mechanism. Recent studies indicate that passage through telomere crisis, with its associated genome instability, plays a significant role in the development of carcinomas (7, 9). It is now critical to determine the role of telomere driven genome instability in the development of liposarcoma and if such a paradigm also holds true for other soft tissue sarcomas.

We thank C. Renner, K. Kaputa, and R. Page of the Fox Chase Cancer Center Histopathology Facility and Fox Chase Cancer Center Tumor Bank; S. Litwin and P. Watts of the Fox Chase Cancer Center Biostatistics Facility; and S. Jablonski of the Fox Chase Cancer Center Cell Imaging Facility.