Whole-Genome Profiling in Liposarcomas Reveals Genetic Alterations Common to Specific Telomere Maintenance Mechanisms

Research Article

Jay E. Johnson,1 Edward J. Gettings,2 Jaclyn Schwalm,2 Jianming Pei,2 Joseph R. Testa,2 Samuel Litwin,2 Margaret von Mehren,2 and Dominique Broccoli1

1Department of Laboratory Oncology Research, Curtis and Elizabeth Anderson Cancer Institute, Memorial Health University Medical Center, Savannah, Georgia and 2Department of Medical Oncology, Population Science Division and Human Genetics Program, Fox Chase Cancer Center, Philadelphia, Pennsylvania

Abstract tional telomere that activates DNA damage checkpoints (2, 3), Telomere attrition ultimately leads to the activation of initiating protective cellular responses, such as apoptosis (4) or protective cellular responses, such as apoptosis or senescence. senescence (5). Although such responses typically limit the like- Impairment of such mechanisms can allow continued prolif- lihood that an aberrantly proliferating cell will cause a tumor, eration despite the presence of dysfunctional telomeres. Under impairment of pathways responsible for the maintenance of such conditions, high levels of genome instability are often genome integrity can allow continued proliferation despite the engendered. Data from both mouse and human model systems presence of dysfunctional telomeres. Under these conditions, high indicate that a period of genome instability might facilitate levels of genome instability are often engendered (6). Interestingly, tumorigenesis. Here, we use a liposarcoma model system to data from both mouse and human model systems indicate that assay telomere maintenance mechanism (TMM)–specific a transient period of genome instability might actually promote genetic alterations. A multiassay approach was used to assess cancer formation (6, 7). the TMMs active in tumors. Genomic DNA from these sam- Activation of a telomere maintenance mechanism (TMM) has ples was then analyzed by high-resolution DNA mapping array been found to stabilize telomeres, facilitate evasion of cell cycle to identify genetic alterations. Our data reveal a higher level checkpoints, and allow the unlimited cellular proliferation that of genome instability in alternative lengthening of telomere is a hallmark of cancer (8, 9). Maintenance of telomeric DNA is (ALT)–positive tumors compared with telomerase-positive usually accomplished through the action of telomerase, a large tumors, whereas tumors lacking both mechanisms have multisubunit complex that adds polynucleotide repeats to pre- relatively low levels of genome instability. The bulk of the existing telomeres (10). In humans, the core telomerase holoen- genetic changes are amplifications, regardless of the mode zyme consists of a catalytic protein subunit (hTERT) and a of telomere maintenance used. We also identified genetic template RNA (hTER). Whereas most tumors use telomerase to changes specific to the ALT mechanism (e.g., deletion of chro- maintain telomeric arrays, sarcomas often use a telomerase- mosome 1q32.2-q44)as well as changes that are underrepre- independent mechanism called alternative lengthening of telo- sented among ALT-positive tumors, such as amplification of meres (ALT; refs. 11, 12). chromosome 12q14.3-q21.2. Taken together, these studies pro- There are several differences between cells that use ALT for vide insight into the molecular pathways involved in the regu- telomere maintenance and those that use telomerase. Whereas lation of ALT and reveal several loci that might be exploited telomerase-positive cells contain relatively short telomeres, telo- either as prognostic markers or targets of chemotherapeutic meric tracts in ALT-positive cells are heterogeneously sized and intervention. [Cancer Res 2007;67(19):9221–8] have a greater mean length than their telomerase-positive counterparts (11). Another characteristic of ALT-positive cells is the Introduction presence of extrachromosomal DNA circles made up of telomeric sequence. ALT-positive cells also contain cell cycle–regulated Telomeres are specialized structures found at the ends of linear colocalizations of telomeric DNA, the telomere binding proteins DNA molecules and are required for the maintenance of eukaryotic TRF1 and TRF2, and the promyelocytic leukemia (PML) nuclear chromosomes. These nucleoprotein structures guard against chro- body in structures called ALT-associated PML bodies (APBs; refs. mosome end-to-end fusions and prevent the natural ends of chro- 13–15). Interestingly, several lines of evidence implicate telomeric mosomes from being recognized as dsDNA breaks (1). Telomeres recombination as playing a role in ALT. For example, telomere suffer sequence attrition each time a cell replicates its genome dynamics in ALT-positive cells are consistent with a recombina- due, in part, to the incomplete nature of semiconservative DNA tion-based mechanism of elongation (16). Furthermore, a tag from replication. Processing events necessary for the maintenance of a single marked telomere is readily transferred to other telo- telomere structure might also contribute to sequence attrition. meres in ALT cells (17). Moreover, several groups have shown Eventually, unchecked loss of telomeric DNA produces a dysfunc- that there is increased telomeric sister chromatid exchange in ALT cells (17–19). Finally, several studies have shown that survivors of telomerase deficiency in yeast maintain their telomeres by a Requests for reprints: Dominique Broccoli, Department of Laboratory Oncology Research, Curtis and Elizabeth Anderson Cancer Institute, Memorial Health University mechanism that is likely functionally similar to ALT in humans and Medical Center, 4700 Waters Avenue, Savannah, GA 31404. Phone: 912-350-0957; that this mechanism proceeds by DNA recombination (20–22). Fax: 912-350-1269; E-mail: [email protected]. I2007 American Association for Cancer Research. Dysfunctional telomeres cause genome instability, whereas both doi:10.1158/0008-5472.CAN-07-1133 ALT and telomerase rescue telomere dysfunction. Several studies www.aacrjournals.org 9221 Cancer Res 2007; 67: (19). October 1, 2007

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Cancer Research have explored the relationship between genome instability and and ALT-negative tumors, we used Fisher’s exact two-sided test. This test telomerase activation, including recent work using a human breast was applied at each SNP locus where the fraction of ALT-positive tumors cancer model system (23). The effect of ALT activation on the levels with a CN change was compared with the fraction of ALT-negative tumors of genome instability in human cancer has not yet been deter- also showing a change. This method was used to analyze amplifications of SNPs within chromosome 12q and deletion of SNPs within chromosome 1q. mined. Here, we use DNA mapping array technology to assay For chromosome 12q, 360 adjacent SNPs were compared. Eight hundred genomic imbalances in a set of liposarcomas, allowing for direct seventy-seven SNPs were compared for chromosome 1q. comparison of the levels of genome instability associated with the We used the Jonckheere-Terpstra test of the null hypothesis that there is various TMMs. This approach has the added benefit of facilitating no relation of the TMM with the levels of either genome or telomere the identification of TMM-specific genetic alterations, analysis instability versus the alternative that instability levels vary with the mode of of which should contribute to our understanding of the ALT and telomere maintenance; specifically, that the instability observed in unknown telomerase mechanisms. tumors is less than that of telomerase-positive tumors, which in turn is less than that of ALT-positive tumors. PCR sequencing of TP53. At least 150 ng of genomic DNA were Materials and Methods subjected to PCR amplification using primer pairs to amplify the regions Tumor specimens. Our tumor sample set consisted of 37 liposarcomas containing exons 2 to 4 (5¶-CCAGGTGACCCAGGGTTGGAAG-3¶ and 5¶-GA- collected between 1992 and 2006 and made available for analysis from AGCCAAAGGGTGAAGAGGAAT-3¶)or5to9(5¶-TTCACTTGTGCCCTGA- the Fox Chase Cancer Center (FCCC) Tumor Bank. Of these samples, 34 CTT-3¶ and 5¶-CTGGAAACTTTCCACTTGAT-3¶). For amplification of exons were described previously with respect to both morphologic characteristics 2 to 4, reactions were done in a buffer containing 16.6 mmol/L ammonium and TMMs (24). In the intervening time, three new samples were obtained acetate, 67 mmol/L Tris-OH(pH8.8), 6.7 Amol/L EDTA (pH8.0), 1% DMSO, and tested for the presence of TMMs. Tumors were obtained from males and 10 mmol/L h-mercaptoethanol, and 3.5 mmol/L magnesium chloride. females with equal frequency, and the median age of the patients was Reactions were incubated at 94jC for 3 min and subjected to two rounds 65 years. Our collection contained mostly grade 1 tumors (n = 27), several each of the following cycles: 94jC for 1 min, 62jC for 1 min, and 72jC for grade 2 tumors (n = 7), and the remainder were grade 3 (n = 3). Eighteen 1.5 min; 94jC for 1 min, 61jC for 1 min, and 72jC for 1.5 min. Next, samples were primary tumors, with the remainder (n = 19) representing reactions were subjected to an additional 33 cycles of amplification using recurrences. Two tumors thought previously to be related were shown by the following conditions: 94jC for 1 min, 60jC for 1 min, and 72jCfor genotyping to actually be two genetically distinct occurrences. The tumor 1.5 min. For amplification of exons 5 to 9, the buffer was formulated as characteristics (Table 1) have been updated given this finding. described above with the exception of containing ammonium sulfate in Assays of telomere maintenance. A multiassay approach was used to place of ammonium acetate and containing 6.7 mmol/L magnesium determine the TMM active, if any, in each tumor. These assays include the chloride. Reactions were incubated at 94jC for 3 min and subjected to two telomeric repeat amplification protocol assay (24, 25), reverse transcription rounds each of the following cycles: 94jC for 1 min, 60jC for 1 min, PCR–based detection of telomerase mRNA (24, 26), telomere restriction and 72jC for 1.5 min; 94jC for 1 min, 59jC for 1 min, and 72jC for 1.5 min. fragment length analysis (24, 25, 27, 28), and indirect immunofluorescent Next, reactions were subjected to an additional 33 cycles of amplification staining of APBs (24) and have been described previously. using the following conditions: 94jC for 1 min, 58jC for 1 min, and 72jC DNA copy number and genotyping assay. Genomic DNA was prepared for 1.5 min. Finally, reaction products were extended for an additional from tumor samples by SDS lysis and phenol extraction and checked for 10 min at 72jC. Sequencing of the resulting PCR products was done at purity using a NanoDrop spectrophotometer. DNA was also checked for either the FCCC Sequencing Facility or at SeqWright (Houston, TX), using excessive shearing by agarose gel electrophoresis. Tumor DNA was then sequencing primers specific to exons 2 to 4 (5¶-CCAGGTGACCCAGGG- analyzed by the Affymetrix GeneChip Single Nucleotide Polymorphism TTGGAAG-3¶ and 5¶-GAAGCCAAAGGGTGAAGAGGAAT-3¶), exon 5 (5¶-TGA- (SNP) 100K mapping assay to identify genomic alterations. This assay has GGAATCAGAGGCCTGG-3¶), exon 6 (5¶-AGAGACGACAGGGCTGGT-3¶), been described previously (29). The intensity of probe signals following exon 7 (5¶-GAGGCAAGCAGAGGCTGG-3¶), exon 8 (5¶-CCTTACTGCCTCT- hybridization to the GeneChip was determined using Affymetrix GeneChip TGCTTC-3¶), or exon 9 (5¶-TTATGCCTCAGATTCACTTTT-3¶). Operating software version 1.2. The Affymetrix Copy Number Analysis Tool version 2.1 was used to generate copy number (CN) calls for each of 57,290 discreet SNP markers, using the default genomic smoothing window setting Results of 0.5 Mb. Both CN and loss of heterozygosity (LOH) calls were generated Analysis of CN changes and LOH. Characterization of telomere using the Affymetrix Dynamic Model. A locus was determined to be maintenance in our liposarcoma sample set revealed that 24% amplified if the reported sample CN was greater than the wild-type (WT) of tumors are ALT-positive (n = 9), 24% are telomerase-positive reference (two copies; 2N) by >0.5N. Similarly, a locus was considered to be (n = 9), and the remainder do not have characteristics consistent deleted if the CN was less than the value of the WT by at least 0.5N. with either mechanism (n = 19). To identify correlations between Genotyping was done using the Affymetrix GeneChip DNA Analysis software version 3.0 with default settings of 0.25 for both homozygous the active TMM and genetic alterations, we did DNA CN and and heterozygous call thresholds. The resultant genotype calls were used to genotyping analyses on DNA obtained from our tumor specimens. determine the likelihood of LOHfor each SNP marker. A locus was A total of 24 tumors were analyzed by DNA mapping array, 4 of considered to show LOHif the probability of its homozygosity correlating which represent recurrences of other tumors in the sample set. with diploidy was 0.01% or less. LOHmeasurements on unpaired samples, Genome instability was measured as the percentage of total SNPs as done here, were achieved by comparing the observed genotypes with assayed (n = 57,290) that deviate from a WT CN. Basal levels of the heterozygosity rates of 110 normal reference samples. The X and Y genome instability were determined by mapping array analysis of chromosomes were omitted from all analyses. normal peripheral blood mononuclear cells obtained from the Data analysis. We used nonparametric statistical tests to avoid same patient as tumor 32. DNA from these control cells showed, as distributional assumptions. Amplifications were compared with deletions expected, a relatively low proportion (2%) of the total SNPs by a two-sided Wilcoxon one-sample test where, for each tumor, the fraction of deleted SNPs was subtracted from the fraction of amplified SNPs. In cases deviating from a WT CN. Studies have shown that CN variable where such differences were all positive, we made use of the simpler two- regions (CNVR) exist (30), wherein certain SNP markers might sided sign test. show CN variation between individuals without being indicative of To compare the extent of amplification or deletion of commonly altered genome instability, per se. Therefore, optimal analysis of these regions (i.e., chromosome 12q and chromosome 1q) between ALT-positive samples would entail comparison of SNP CNs observed in tumor

Cancer Res 2007; 67: (19). October 1, 2007 9222 www.aacrjournals.org

Table 1. Telomere maintenance and genome instability in liposarcomas

c b c Sample Grade TMM GI* GI Amp:Del TI TI Amp:Del LOH

Tumor 1 1 ALT 22 18:4 30 24:6 76 Tumor 2 1 ALT 3 2:1 6 5:1 8 Tumor 3 1 ALT 5 4:1 8 7:1 17 Tumor 4 1 ALT 33 19:14 49 32:17 17 Tumor 5 1 ALT 33 20:13 38 26:12 44 Tumor 6 1 ALT ND ND ND ND ND Tumor 7 1 ALT 32 18:14 46 28:18 41 Tumor 8 1 UNK 25 9:16 29 9:20 29 Tumor 9 1 TELALT 6 4:2 6 5:1 17 Tumor 10 1 UNK 1 1:0 2 2:0 15 Tumor 11 1 UNK 4 3:1 4 3:1 16 Tumor 12 1 UNK ND ND ND ND ND Tumor 13 1 UNK ND ND ND ND ND Tumor 14 1 UNK ND ND ND ND ND Tumor 15 1 TEL 19 11:8 22 17:5 21 Tumor 16 1 UNK ND ND ND ND ND Tumor 17 1 UNK ND ND ND ND ND Tumor 18 1 UNK ND ND ND ND ND Tumor 19 1 UNK ND ND ND ND ND Tumor 20 1 TEL 1 1:0 3 3:0 13 Tumor 21 1 TEL ND ND ND ND ND Tumor 22 2 ALT ND ND ND ND ND Tumor 23 2 TEL 1 1:0 1 1:0 13 Tumor 24 2 UNK ND ND ND ND ND

Tumor 25 2 UNK ND ND ND ND ND Tumor 25R1 1 UNK ND ND ND ND ND

Tumor 26 1 UNK 3 3:0 5 4:1 15 Tumor 26R1 2 UNK 7 6:1 14 13:1 16 Tumor 26R2 2 TEL 15 8:7 19 15:4 16

Tumor 27 1 TEL 12 6:6 13 5:8 19 Tumor 27R1 2 TEL 18 10:8 25 14:11 19 Tumor 27R2 2 TEL 20 10:10 18 11:7 23

Tumor 28 1 UNK 3 3:0 2 2:0 15 Tumor 29 1 UNK 2 1:1 2 1:1 15 Tumor 30 3 ALT 25 19:6 30 20:10 12 Tumor 31 3 UNK 9 6:3 14 9:5 16 Tumor 32 3 UNKTEL 19 14:5 25 17:8 29

Normal NA ND 2 2:0 5 5:0 14

Abbreviations: NA, not applicable; ND, not done; TEL, telomerase; UNK, unknown. *Percentage genome instability. cPercentage instability composed of amplifications and deletions, respectively. bPercentage telomere instability.

samples with those observed in normal DNA obtained from the to CN variation. Our results indicate that ALT-positive lip- same individual. Because matched normal DNA was not available osarcomas have, on average, more CN changes than telomerase- for the majority of the tumors interrogated here, we instead positive tumors. Furthermore, telomerase-positive tumors have compared tumor CN changes against the CNVRs published in the more CN changes than tumors with no evidence of telomere Database of Genomic Variants3 and determined that the genetic maintenance (ALT>telomerase>unknown; P = 0.031; Fig. 1A). The alterations observed in the liposarcoma tumors are likely not due majority of ALT tumors (five of seven) exhibited >20% of SNPs deviating from diploid with an average of 21.9% of SNPs having altered CN. Telomerase-positive tumors had on average 11.2% of 3 http://projects.tcag.ca/variation/ SNPs deviating from a WT CN, with three of the five tumors www.aacrjournals.org 9223 Cancer Res 2007; 67: (19). October 1, 2007

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Cancer Research exhibiting 15% to 20% of SNPs with CN alterations. None of the sample, itself. If these tumors were to be excluded from analysis, the telomerase-positive tumors had >20% of SNPs exhibiting CN remaining tumors without evidence of telomere maintenance changes. Although relatively few (6%) of the SNPs deviated from a would exhibit CN changes in only 4.3% of SNPs, on average. WT CN in tumor 9, this sample also showed both telomerase Telomere dysfunction drives genome instability. Accordingly, it is activity and an intermediate APB phenotype, possibly indicating possible that alterations in CN at telomere-proximal loci might the activation of both telomere maintenance pathways within precede changes at other genetic loci, producing a higher frequency this tumor. Finally, the majority of tumors without evidence of of CN changes at peritelomeric regions. Thus, instability at telomere maintenance (six of nine) had <10% of SNPs deviating telomere-proximal regions was also measured, only considering from a WT CN with an average of only 7.3% of SNPs exhibiting CN SNPs positioned within the terminal 10 Mb of each chromosome. changes. One of two tumors in this category (tumor 8) that Similar to the results for overall instability, the levels of peri- exhibited a high frequency of CN changes (25%) also had an telomeric CN change were higher in ALT-positive tumors than in intermediate APB phenotype (data not shown), possibly indicating their telomerase-positive counterparts, which in turn showed more that this tumor had a weak ALT phenotype or was composed of a CN changes than those tumors lacking evidence of both mecha- heterogeneous population of cells, some of which were ALT positive. nisms (P = 0.0128). Interestingly, instability near the telomere was Cultured cells from the other lesion (tumor 32, 19% CN changes) typically higher than that for the genome as a whole (Fig. 1B). showed transient telomerase activity (data not shown), suggesting Analysis of tumor genomic DNA by mapping array also allows the that this mechanism might have been active sporadically during the identification of SNP markers that show a high likelihood of LOH. development of this lesion, despite not being detected in the tumor Our results indicate that LOHis more prevalent in ALT-positive tumors than in those lesions that use telomerase or have no evidence of telomere maintenance (Fig. 1C). On average, 38% of SNPs in ALT-positive tumors showed evidence of LOH, with the majority of tumors (five of seven) showing levels much greater than what was observed for the normal peripheral blood control (14%). SNPs (16.4%) exhibit LOHin the average telomerase-positive liposarcoma (n = 5), a value similar to that of the WT control. The same is true of those tumors without evidence of telomere maintenance, which show LOHat an average of only 18.8% of SNPs. Only two of these samples exhibit LOHat a substantially greater frequency than the WT control, tumors 8 and 32 (29% of SNPs, each). As discussed above, it is possible that these two lesions exhibited either low-level ALT (tumor 8) or transient telomerase activity (tumor 32). Excluding these samples from analysis reveals LOHat an average of 15.3% of SNPs in the remaining tumors. The data indicate that ALT utilization is associated with relatively high levels of LOH. In addition, LOH does not seem to simply be a function of the tendency toward genetic rearrangements, as there is little difference between the extent of LOHin telomerase-positive tumors (intermediate levels of genome instability) and those tumors without detectable telomere maintenance (low levels of genome instability). The majority of CN changes are amplifications. Regardless of the category of tumor analyzed (ALT, telomerase, and unknown), CN changes tend to represent amplifications rather than deletions (Fig. 1; Tables 1 and 2). This tendency toward amplifications is seen both at interstitial regions of the genome and at telomeres and is most striking for ALT-positive tumors (overall and peritelomeric; P = 0.0156). In addition, a similar but less significant tendency is observed in telomerase-positive lesions (P = 0.0625 and 0.0431, respectively) and those tumors lacking both mechanisms (P = 0.0313 and 0.236, respectively). In the most extreme case (analysis of overall genome instability in tumors with no evidence of ALT or telomerase), f80% of all markers that deviate from a WT CN are amplified. The least relative amplification is seen by analysis of overall genome instability in telomerase-positive tumors, wherein f60% of all non-2N markers represent amplifications. Although these numbers represent averages obtained from analysis of all tumors in each category, the propensity toward amplifications is also observed in individual tumor samples (Table 1). In only 2 (9%) of the 24 tumors analyzed (tumors 8 and 26) do the frequency of deletions exceed that of amplifications. For tumor 26, this is true Figure 1. Levels of genome instability and LOH in liposarcomas. Stacked bar graphs showing the relative levels of overall genome instability (A), peritelomeric only if the CN changes at telomere-proximal loci. In tumor 8, genome instability (B), and LOH (C) in liposarcoma tumors. however, the number of deletions exceeds that of amplifications in

Cancer Res 2007; 67: (19). October 1, 2007 9224 www.aacrjournals.org

Table 2. Frequency of amplifications and deletions in liposarcomas

TMM Total genome Telomeric 10 Mb

Amplifications Deletions Amplifications Deletions

% All loci % CN changes % All loci % CN changes % All loci % CN changes % All loci % CN changes

ALT 14.1 69.2 27.5 30.8 20.4 73.9 9.2 26.1 Telomerase 6.3 61.9 4.9 38.1 9.3 77.6 3.3 22.4 Unknown 5.2 78.9 1.6 21.1 6.9 75.9 2.3 24.1 both interstitial and telomeric regions. Interestingly, tumor 8 also respect to TMM. The experiments done here allow both a higher represents the only tumor with no strong evidence of ALT or resolution analysis of CN changes and identification of any changes telomerase that also exhibited relatively high levels of genome correlated with a specific subset of tumors. The most common instability (25% overall and 29% peritelomeric). As mentioned alteration observed in liposarcomas, amplification of an interstitial previously, this tumor showed an APB phenotype intermediate to region of chromosome 12q, is also the most common alteration in those tumors that use ALT and those that do not, which, in our sample set. Based on our results, chromosome 12q also connection with its high level of genome instability, makes it represents the portion of the genome featuring the highest average atypical in the context of the sample set. CNs in all liposarcomas, ranging from four to eight copies. Our Analysis of telomere maintenance, genome instability, and analyses have identified the minimal common amplicon as being tumor progression. In a recent study, positive correlations were chromosome 12q14.3-q21.2 (Fig. 3A) and showed that this identified between activation of telomere maintenance and tumor amplification is underrepresented among ALT-positive tumors. grade in liposarcomas. The authors found that the majority Interestingly, the CDK4 gene, which has been shown previously to (f70%) of grade 1 lesions showed no evidence of ALT or telo- be amplified in soft tissue sarcomas (32), is not part of the merase (31). In grade 2 and 3 lesions, however, the converse was contiguous chromosome 12q14.3-q21.2 amplicon identified in this true, with f70% of tumors showing evidence of a TMM. We sought study. Nevertheless, it is similarly amplified in at least half of to extend these studies by considering possible associations tumors lacking ALT and in only one ALT-positive sample (14%). between telomere maintenance and tumor progression in our Additionally, we identified deletion of chromosome 1q32.2-q44 as sample set. For grade 1 lesions, tumors that were TMM positive being a genetic change frequent in tumors that use the ALT (n = 12, 46%) and tumors that lacked both ALT and telomerase (n = 14, 54%) were present in comparable numbers. In grade 2 and 3 lesions, similar frequencies were observed, with exactly equal numbers of tumors showing telomere maintenance as those that did not (n = 5, each). Therefore, in contrast with previous studies, our data show no particular correlation between tumor grade and TMM and are consistent with the activation of a telomere maintenance being an early event in the development of liposarcomas. In consideration of recurrent tumors, telomere attrition correlates with increased instability of both the telomere and the whole genome. For tumor 26, each successive recurrence shows progressively shorter average telomere lengths compared with the primary tumor (Fig. 2). In terms of the associated levels of genome instability, an inverse proportionality is indicated, with telomere and overall instability increasing as average telomere length decreases. Of these three samples, telomerase activity is detected only in the last occurrence. With respect to tumor 27 and its recurrences, a reduction in average telomere length was observed from the primary tumor to the first recurrence, whereas there is relatively little increase in either telomere or overall genome instability. Furthermore, there is relatively little telomere shortening from the first recurrence to the second, presumably due to the action of telomerase, and the levels of peritelomeric and overall instability are similarly maintained. Identification of TMM-specific genetic changes by whole- genome profiling. Previous studies assessing genetic changes in soft tissue sarcomas, and specifically in liposarcomas, using Figure 2. Telomere maintenance and genome instability in recurrent tumors. genome-wide approaches have relied largely on comparative Recurrent tumor sets derived from tumor 26 (A) and tumor 27 (B) were analyzed. Average telomere lengths (black). Levels of instability (gray). Tumors in which genomic hybridization using bacterial artificial chromosome arrays telomerase activity was detectable (italics). The remaining two tumors and also failed to stratify tumors, or their associated changes, with showed no evidence of telomerase or ALT. www.aacrjournals.org 9225 Cancer Res 2007; 67: (19). October 1, 2007

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Cancer Research mechanism (Fig. 3B). This is a large deletion spanning over 40 Mb tion might cause a more permissive environment for the activation and encompassing several genes whose functions make them of the ALT mechanism (34). Apparently, such inhibition might also candidate regulators of ALT. For example, the deletion spans the result in the generation of a potentially tumor-promoting unstable gene encoding the p53-binding protein 2 (TP53BP2/ASPP). Binding genetic milieu. of 53BP2 to p53 competes with the association of the latter factor To assess TP53 status, we did PCR sequencing of exons 2 to 9, with promoter regions involved in the activation of cellular which would contain the majority of all reported mutations should senescence, thereby sensitizing cells to p53-dependent apoptotic any be present in these tumors. The sequence of TP53 was inter- death. As discussed below, it is possible that impairment of p53 rogated in a total of 24 tumors (ALT, n = 7; telomerase, n =7; might facilitate ALT activation. Additional candidate genes that unknown, n = 10). The WT status of TP53 was confirmed in all of might play a role in the regulation of ALT, telomerase, or the lesions interrogated, with the exception of tumor 2, which tumorigenesis, in general, are summarized in Table 3. showed a missense mutation at codon 60 (CCA>GCA). This The TP53 DNA surveillance gene is WT in liposarcomas. mutation has not been reported previously. Sequence analyses have Considered the ‘‘guardian of the genome,’’ the TP53 gene is mutated also identified several commonly observed polymorphisms, includ- in nearly half of all cancers. In a study assaying p53 status in soft ing the well-characterized codon 72 (proline/arginine) polymor- tissue sarcomas, it was found that there was a high incidence of phism.4 The proline and arginine variants are present in 45% and TP53 mutations in tumors of histologic grade 3 (33). In addition, 55%, respectively, of nonrecurrent liposarcoma tumor samples our laboratory and others have suggested previously that (data not shown); these frequencies are not appreciably different impairment of the activity of p53 in suppressing DNA recombina- from what is observed in the U.S. population as a whole. Detection of such changes shows the efficacy of this approach in identifying sequence alterations when present. Our data indicate that mutation of the TP53 DNA surveillance gene does not play a significant role in the activation of ALT or the accrual of genome instability in liposarcomas. As mentioned above, previous work has shown that mutation of TP53 seems to correlate with high histologic grade, whereas our sample set contains mostly low and intermediate grade lesions. Thus, although our data suggest that mutation of TP53 is not required for the initial development of liposarcomas, it remains possible that TP53 inactivation might play a role in the evolution of more advanced malignancies. Discussion DNA mapping analyses have revealed that ALT-positive tumors have, on average, higher levels of genome instability than do their telomerase-positive counterparts. This finding suggests that, compared with telomerase, ALT might be less efficient at stabilizing telomere-driven genome instability or activated following a longer period of telomere dysfunction. Another possibility is that the spectrum of mutations that causes ALT also results in high levels of genome instability independent of telomere status. This possibility is intriguing given the likelihood that the ALT mechanism functions by DNA recombination. Our sample set contained tumors that showed no evidence of telomere maintenance, analysis of which showed that the majority of these lesions had relatively few CN changes. These findings are consistent with previous reports that neither activation of a TMM nor telomere-driven genome instability is obligatory steps in the formation of a clinically detectable lesion. The fact that tumors lacking both known TMMs typically have relatively low levels of genome instability suggests that these lesions have not yet undergone a telomere crisis event. Theoretically, this is possible if the progenitor cells that cause such tumors possess sufficient telomeric reserves to overcome the proliferation-induced telomeric shortening that is usually associated with tumorigenesis. Indeed, it has been suggested that under conditions of low cell turnover, transformed cells can produce a clinically significant and aggressive tumor within relatively few population doublings (35). Figure 3. High-resolution analysis of chromosome 12q amplifications and 1q deletions as a function of TMM in liposarcomas. Stacked area graphs showing Analysis of genotyping data indicates that LOHis prevalent in the proportion of tumors that show amplification or deletion of the indicated liposarcomas that use ALT. Furthermore, the observed LOHdoes regions of chromosome 12q (A) and chromosome 1q (B). Regions wherein the CNs in ALT-positive and ALT-negative lesions vary significantly from one another (P < 0.05; tall red columns). Regions wherein such differences approach significance (0.10 > P > 0.05; narrow red columns). 4 http://p53.bii.a-star.edu.sg/index.php

Cancer Res 2007; 67: (19). October 1, 2007 9226 www.aacrjournals.org

Table 3. TMM-specific genetic alterations of chromosome 1q and chromosome 12q identified by whole-genome profiling

Chromosome Band Gene Physical position Description

1 q32.1 RBBP5 203322602-203357754 Regulator of proliferation through interactions with hypophosphorylated retinoblastoma protein 1 q32.1 ELK4 203855021-203868623 Member of the ETS family of transcription factors, involved in c-fos regulation 1 q32.1 RAB7L1 204004301-204010747 A RAS-related factor homologous to RAB7, a member of the small GTPase RAB subfamily 1 q42.13 TAF5L 227801565-227828417 A member of the PCAF histone acetylase complex, which is required for myogenic differentiation 12 q14.3 HMGA2 64504507-64646335 Alterations of this gene are found in myxoid liposarcomas 12 q14.3 HELB 64982623-65018223 A DNA helicase involved in DNA replication and cell cycle regulation 12 q15 DYRK2 66329021-66340408 Involved in cellular growth and/or development, as well as phosphorylation of histones H3 and H2B in vitro 12 q15 MDM2 67488247-67520481 Its overexpression results in degradation of p53 and the abrogation of its tumor suppressor function 12 q15 YEATS4 68039799-68070842 A transcription factor shown to be amplified in tumors 12 q15 PTPRB 69271781-69317487 Potentially involved in such processes as growth, differentiation, mitosis, and oncogenic transformation 12 q15 PTPRR 69318129-69434632 Potentially involved in such processes as growth, differentiation, mitosis, and oncogenic transformation 12 q21.1 TSPAN8 69805144-69838046 Encodes a cell surface glycoprotein known to complex with integrins and is expressed in varying carcinomas 12 q21.1 THAP2 70344051-70360684 Contains a THAP domain and is associated with apoptotic signaling 12 q21.2 PHLDA1 74705495-74711823 May play an important role in the antiapoptotic effects of insulin-like growth factor-1

NOTE: The genes listed here represent only those loci that are amplified or deleted in 20% or greater of all liposarcomas assayed. All loci present on chromosome 1q are deleted predominantly in ALT-positive samples, whereas those loci on chromosome 12q are amplified mostly in those samples that lack the ALT mechanism.

not correlate with the level of CN changes, as the extent of LOHin result of multiple mechanisms affecting the stability of the genome. telomerase-positive tumors (11.2% SNP CN changes) was similar to Additional experiments will be necessary to determine the actual what was observed in those tumors without detectable telomere causes of genome instability in liposarcomas, as well as their maintenance (4.3% SNP CN changes). Therefore, it is probable that respective contributions to this phenomenon. the LOHobserved in ALT-positive liposarcomas is either due to With the exception of tumor 8, all liposarcoma tumors show a activities associated with activation of the ALT mechanism or a greater proportion of amplifications than deletions. Our data consequence of the ALT mechanism, itself. Currently, the data are suggest that tumor 8 might have either been heterogeneous with insufficient to distinguish between these possibilities. respect to ALT activation or simply displayed a weak ALT pheno- The observation that instability at telomere-proximal sites is type. However, the molecular mechanisms underlying the unique consistently higher than at interstitial sites of the genome raises the pattern of genome instability observed in this tumor (deletions possibility that the observed instability might be driven by the exceeding amplifications) remain unclear. Future identification breakage-fusion bridge cycle associated with dysfunctional telo- and analysis of other samples that display similar characteristics meres. This observation might also be due, however, to preferential should help to explain this observation. breakage at telomere-proximal fragile sites compared with those Taken together, our data indicate that ALT-positive and ALT- located interstitially (36, 37). Alternatively, abrogation of p53 negative tumors are fundamentally different biological entities, function might result in the genome instability observed in our with ALT-positive tumors showing significantly higher levels of panel of liposarcomas. Although sequence analyses have shown that genome instability and LOHthan their ALT-negative counterparts. the TP53 gene is WT in nearly all tumors assayed, the possibility Moreover, whole-genome profiling has revealed two sizable genetic remains that TP53 function is impaired indirectly. Indeed, ampli- alterations further distinguishing ALT-positive and ALT-negative fication of the MDM2 gene is observed in more than half of lesions. Whereas amplification of chromosome 12q has been iden- liposarcomas and its overexpression has been shown to impair p53 tified previously as the most common genetic change in liposar- function by targeting the protein for ubiquitin-dependent degrada- comas, here we show that amplification of the minimum region, tion (38). However, amplification of MDM2 is observed infrequently chromosome 12q14.3-q21.2, is common only to telomerase-positive in ALT-positive tumors, which also have the highest average levels tumors and those tumors that lack evidence of telomere main- of genome instability. Therefore, in these lesions, impairment of p53 tenance. In addition, deletion of chromosome 1q32.2-q44 is a might be achieved by mutation of one of its downstream effectors. frequent event only in ALT-positive samples. A recent study It is also probable that the observed instability is the combined reported that both ALT and telomerase positivity are associated www.aacrjournals.org 9227 Cancer Res 2007; 67: (19). October 1, 2007

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Cancer Research with unfavorable disease outcome in liposarcomas, although prog- consequently, are likely to display differential sensitivity to DNA- nosis is poorest for patients whose tumors use ALT (31). Ap- damaging agents. Conversely, ALT-positive tumors should prove parently, differences in TMM-specific genetic alterations, such as refractory to treatment with intervention strategies that specifically those shown in the current study, might underlie the observed inhibit telomerase. Rapid identification of the active TMM in differences in patient prognosis. Analysis of candidate factors liposarcomas would allow treatment to be tailored to the indi- encoded within the altered regions might aid development of novel vidual, achieving the highest possible tumor cell killing with the anticancer therapeutics that specifically target the ALT or telo- lowest possible patient toxicity. merase mechanisms. Additionally, these regions might also encode genes that, while not playing a role in telomere maintenance per se, Acknowledgments are tightly linked to either of the specific TMMs. Given the positive Received 3/26/2007; revised 7/23/2007; accepted 7/26/2007. correlations of these mechanisms with tumor aggressiveness, such Grant support: NIHgrants CA117675-01 (J.E. Johnson) and CA098087-03 (D. factors might represent excellent biomarkers for the rapid Broccoli). The costs of publication of this article were defrayed in part by the payment of page identification of the active TMM and, consequently, could serve charges. This article must therefore be hereby marked advertisement in accordance as efficient indicators of prognosis. Finally, identification of ALT- or with 18 U.S.C. Section 1734 solely to indicate this fact. telomerase-associated factors might also serve as predictors of We thank the members of the Broccoli Lab for helpful discussion; P. Cairns of FCCC for assistance with TP53 sequencing; FCCC Sequencing Facility; C. Renner, tumor drug response. For example, ALT-positive tumors show a K. Kaputa, and R. Page of the FCCC Histopathology Center and Tumor Bank; and higher level of genome instability than tumors lacking ALT and, S. Jablonski of the FCCC Cell Imaging Facility.

References in cells in the G2/M phase of the cell cycle. J Cell Sci 27. Grobelny JV, Kulp-McEliece M, Broccoli D. Effects 2000;113:4577–85. of reconstitution of telomerase activity on telomere 1. Ferreira MG, Miller KM, Cooper JP. Indecent exposure: 15. Wu G, Lee WH, Chen PL. NBS1 and TRF1 colocalize maintenance by the alternative lengthening to telo- when telomeres become uncapped. Mol Cell 2004;13: at promyelocytic leukemia bodies during late S/G2 meres (ALT) pathway. Hum Mol Genet 2001;10:1953–61. 7–18. phases in immortalized telomerase-negative cells. Im- 28. Grant JD, Broccoli D, Muquit M, Manion FJ, Tisdall 2. d’Adda di Fagagna F, Reaper PM, Clay-Farrace L, et al. plication of NBS1 in alternative lengthening of telo- J, Ochs MF. Telometric: a tool providing simplified, A DNA damage checkpoint response in telomere- meres. J Biol Chem 2000;275:30618–22. reproducible measurements of telomeric DNA from initiated senescence. Nature 2003;426:194–8. 16. Murnane JP, Sabatier L, Marder BA, Morgan WF. constant field agarose gels. Biotechniques 2001;31: 3. Takai H, Smogorzewska A, de Lange T. DNA dam- Telomere dynamics in an immortal human cell line. 1314–6, 8. age foci at dysfunctional telomeres. Curr Biol 2003;13: EMBO J 1994;13:4953–62. 29. Pei J, Kruger WD, Testa JR. High-resolution analysis 1549–56. 17. Bailey SM, Brenneman MA, Goodwin EH. Frequent of 9p loss in human cancer cells using single nucleotide 4. Karlseder J, Broccoli D, Dai Y, Hardy S, de Lange T. recombination in telomeric DNA may extend the polymorphism-based mapping arrays. Cancer Genet p53- and ATM-dependent apoptosis induced by telo- proliferative life of telomerase-negative cells. Nucleic Cytogenet 2006;170:65–8. meres lacking TRF2. Science 1999;283:1321–5. Acids Res 2004;32:3743–51. 30. Redon R, Ishikawa S, Fitch KR, et al. Global variation 5. van Steensel B, Smogorzewska A, de Lange T. TRF2 18. Bechter OE, Zou Y, Walker W, Wright WE, Shay JW. in copy number in the human genome. Nature 2006;444: protects human telomeres from end-to-end fusions. Cell Telomeric recombination in mismatch repair deficient 444–54. 1998;92:401–13. human colon cancer cells after telomerase inhibition. 31. Costa A, Daidone MG, Daprai L, et al. Telomere 6. Chin L, Artandi SE, Shen Q, et al. p53 deficiency Cancer Res 2004;64:3444–51. maintenance mechanisms in liposarcomas: association rescues the adverse effects of telomere loss and 19. Londono-Vallejo JA, Der-Sarkissian H, Cazes L, with histologic subtypes and disease progression. cooperates with telomere dysfunction to accelerate Bacchetti S, Reddel RR. Alternative lengthening of Cancer Res 2006;66:8918–24. carcinogenesis. Cell 1999;97:527–38. telomeres is characterized by high rates of telomeric 32. Brennan M, Singer S, Maki R, O’Sullivan B. Sarcomas 7. Artandi SE, DePinho RA. A critical role for telomeres exchange. Cancer Res 2004;64:2324–7. of the soft tissues and bone. In: De Vita V, Hellman S, in suppressing and facilitating carcinogenesis. Curr 20. Le S, Moore JK, Haber JE, Greider CW. RAD50 and Rosenberg S, editors. Cancer: principles & practice of Opin Genet Dev 2000;10:39–46. RAD51 define two pathways that collaborate to oncology. Philadelphia (PA): Lippincott Williams & 8. Bodnar AG, Ouellette M, Frolkis M, et al. Extension of maintain telomeres in the absence of telomerase. Wilkins; 2005. p. 1581–686. life-span by introduction of telomerase into normal Genetics 1999;152:143–52. 33. Rieske P, Bartkowiak JK, Szadowska AM, Olborski B, human cells. Science 1998;279:349–52. 21. Lundblad V, Blackburn EH. An alternative pathway Harezga-Bal B, Debiec-Rychter M. A comparative study 9. Vaziri H, Benchimol S. Reconstitution of telomerase for yeast telomere maintenance rescues est1—senes- of P53/MDM2 genes alterations and P53/MDM2 pro- activity in normal human cells leads to elongation of cence. Cell 1993;73:347–60. teins immunoreactivity in soft-tissue sarcomas. J Exp telomeres and extended replicative life span. Curr Biol 22. Teng SC, Zakian VA. Telomere-telomere recombina- Clin Cancer Res 1999;18:403–16. 1998;8:279–82. tion is an efficient bypass pathway for telomere 34. Razak ZR, Varkonyi RJ, Kulp-McEliece M, et al. p53 10. Collins K, Mitchell JR. Telomerase in the human maintenance in Saccharomyces cerevisiae. Mol Cell Biol differentially inhibits cell growth depending on the organism. Oncogene 2002;21:564–79. 1999;19:8083–93. mechanism of telomere maintenance. Mol Cell Biol 11. Bryan TM, Englezou A, Gupta J, Bacchetti S, Reddel 23. Chin K, de Solorzano CO, Knowles D, et al. In situ 2004;24:5967–77. RR. Telomere elongation in immortal human cells analyses of genome instability in breast cancer. Nat 35. Matthews P, Jones CJ, Skinner J, Haughton M, de without detectable telomerase activity. EMBO J 1995; Genet 2004;36:984–8. Micco C, Wynford-Thomas D. Telomerase activity and 14:4240–8. 24. Johnson JE, Varkonyi RJ, Schwalm J, et al. Multiple telomere length in thyroid neoplasia: biological and 12. Bryan TM, Englezou A, Dalla-Pozza L, Dunham MA, mechanisms of telomere maintenance exist in lip- clinical implications. J Pathol 2001;194:183–93. Reddel RR. Evidence for an alternative mechanism for osarcomas. Clin Cancer Res 2005;11:5347–55. 36. Hellman A, Zlotorynski E, Scherer SW, et al. A role for maintaining telomere length in human tumors and 25. Broccoli D, Godley LA, Donehower LA, Varmus HE, common fragile site induction in amplification of tumor-derived cell lines. Nat Med 1997;3:1271–4. de Lange T. Telomerase activation in mouse mammary human oncogenes. Cancer Cell 2002;1:89–97. 13. Yeager T, Neumann A, Englezou A, Huschtscha L, tumors: lack of detectable telomere shortening and 37. Richards RI. Fragile and unstable chromosomes in Noble J, Reddel R. Telomerase-negative immortalized evidence for regulation of telomerase RNA with cell cancer: causes and consequences. Trends Genet 2001;17: human cells contain a novel type of promyelocytic proliferation. Mol Cell Biol 1996;16:3765–72. 339–45. leukemia (PML) body. Cancer Res 1999;59:4175–9. 26. Keith WN, Hoare SF. Detection of telomerase hTERT 38. Toledo F, Wahl GM. Regulating the p53 pathway: 14. Grobelny JV, Godwin AK, Broccoli D. ALT-associated gene expression and its splice variants by RT-PCR. in vitro hypotheses, in vivo veritas. Nat Rev Cancer 2006; PML bodies are present in viable cells and are enriched Methods Mol Med 2004;97:297–309. 6:909–23.

Cancer Res 2007; 67: (19). October 1, 2007 9228 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Whole-Genome Profiling in Liposarcomas Reveals Genetic Alterations Common to Specific Telomere Maintenance Mechanisms

Jay E. Johnson, Edward J. Gettings, Jaclyn Schwalm, et al.

Cancer Res 2007;67:9221-9228.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/67/19/9221

Cited articles This article cites 37 articles, 13 of which you can access for free at: http://cancerres.aacrjournals.org/content/67/19/9221.full#ref-list-1

Citing articles This article has been cited by 5 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/67/19/9221.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/67/19/9221. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.