Prevalence and potentially prognostic value of C‐circles associated with alternative lengthening of telomeres in canine appendicular osteosarcoma

Abstract Alternative lengthening of telomeres (ALT) is a telomerase‐independent telomere maintenance mechanism (TMM) with high prevalence in human osteosarcomas but remains unknown in canine osteosarcomas. The aim of this study was to evaluate the prevalence of ALT by detection of extra‐chromosomal circles of telomeric DNA and to assess clinical outcome in canine patients with spontaneous occurring appendicular osteosarcoma. Fifty dogs with histopathological confirmed osteosarcomas were included into this study. Medical records were retrospectively analysed for patient characteristics, oncologic therapy and survival. DNA was isolated from archived FFPE tumour tissue specimens and applied for C‐ and G‐circle assay (CCA and GCA) and for telomeric content (TC) measurement with radiolabeled probes. ALT activity was detected for 10 of 50 (20%) cases by CCA. Four CCA positive cases were detected even with input DNA below 1 ng and demonstrated the high sensitivity of CCA for canine tumours. G‐circles and TC were not suitable to distinguish CCA positive and negative cases. CCA‐status showed an association with male gender and Rottweiler breed. Dogs with CCA positive osteosarcomas had shorter overall survival times than patients with CCA‐tumours and CCA‐status was a significant prognostic factor besides treatment in the Cox proportional hazard model. These findings make canine osteosarcomas an interesting model for comparative TMM research, but future studies are warranted to investigate if CCA‐status can serve as novel prognostic marker.

tandem DNA hexanucleotide repeats (TTAGGG)n in a multiplicity length. 2,3 During the life span of normal somatic cells, telomeres shorten progressively and lead to replicative senescence or apoptotic cell death. 1 In cancer cells, activation of a telomere maintenance mechanism (TMM) is an essential step for immortalization. 4 In cancer, there are two known TMMs, telomerase activity (TA) and alternative lengthening of telomeres (ALT). The majority of human and canine cancer types activate the enzyme telomerase, as TMM, 3,5 whereas 4% to 11% of human tumours use the ALT-pathway. 6 In a recent largescale study across 31 human cancer types telomerase was activated in 73%, ALT in 5% and no TMM activation was observed in 22%. [7][8][9] Comparable to humans, several studies reported that more than 90% of canine tumours utilize telomerase as TMM. [10][11][12][13] In contrast, information about ALT in canine tumours is currently limited. 14 ALT is a telomerase-independent TMM based on homologous recombination with potential as a quantifiable clinical marker and as a target for cancer therapy. 6,15 For the detection of ALT certain characteristics have been used in humans including long and heterogeneous telomere length, colocalization of telomeres with ALT-associated PML bodies, mutation/loss of ATRX/DAXX or presence of C-circles. [16][17][18][19] In dogs, ALT-associated characteristics, such as long heterogeneous telomeres, overexpression of p53, colocalization of DAXX with telomeres and presence of C-circles, have been described. 14 C-circles are extra-chromosomal circles of telomeric DNA that are specific for ALT activity. 15 The detection of C-circles has developed to a front-line assay for identification of ALT activity. 16,19,20 The C-circle assay (CCA) involves extraction of DNA, amplification of self-primed C-circles with rolling circle amplification (RCA) and detection of the amplification products by native telomeric dot blot or telomeric qPCR. 19 Besides C-circles, G-circles are ALT specific and have the ability to self-prime RCA of telomeric G-strand templates, but they are about 100-fold less abundant. 15,16 The CCA is the first ALT assay that has a quantitative correlation with ALT activity levels and can be used for liquid biopsies as well as in FFPE archived human and canine tissue. 14,15,21 Certain tumour types seem to utilize ALT more frequently. Especially, in tumours originating from mesenchymal tissues, such as sarcomas, the ALT phenotype is overrepresented. [7][8][9] Sarcomas are a heterogeneous cancer group and ALT status has been demonstrated as negative prognostic factor in some sarcoma subtypes. [22][23][24][25] In dogs, the ALT mechanism has been identified in various sarcoma subtypes and as in humans different sarcoma subtypes seem to have a heterogeneous ALT prevalence. 14 In human osteosarcomas, high ALT prevalence rates from 47% to 80% are observed. [26][27][28][29] Osteosarcoma (OSA) is a highly malignant bone tumour in humans and dogs, characterized by a high metastatic rate and low overall survival rates. In both species, it represents the most common primary malignant skeletal neoplasia. 4,[30][31][32] The annual incidence in people is reported with approximately one to five cases per million population. In comparison to humans, OSA incidence rates in dogs are about 27 times higher. [32][33][34] This high incidence rate would make canine osteosarcomas an interesting model for comparative TMM research. However, the clinical relevance of ALT in canine osteosarcomas is currently unknown. Our previous study in various sarcomas included four osteosarcomas and one showed ALT activity. 14 Thus, the aim of this study was to evaluate the prevalence of ALT in a larger cohort of canine appendicular osteosarcomas together with clinical outcome.

| Patient cohort and clinical data
Dogs with available archived formalin-fixed and paraffin embedded (FFPE) appendicular osteosarcoma samples that had been collected during routine diagnostic work-up and therapy at our institution between 2003 and 2016 were included into this study. Medical records were analysed for the following information: Patient signalment, tumour location and tumour subtype, disease stage, pretreatment alkaline phosphatase (ALP) level, therapy and outcome.
Routine staging procedures included radiographs of the primary lesion, three-view thoracic radiographs, CBC and biochemistry profile.
In patients with hind limb lesions an abdominal ultrasound was performed. Additional staging procedures were only obtained if clinical findings or routine staging results required further diagnostic workup. Dogs were assigned to the age categories "young" (<5 years), "middle-aged (5-10 years) and "old (>10 years). Further, patients were assigned to the three size categories "small-medium" (< 30 kg), "large" (30-60 kg) and "giant" (> 60 kg). Median overall survival time (OST) was defined as the time from diagnosis until death.

| Spontaneous OSA tissue samples and controls
Fifty archived FFPE tumour tissue samples from spontaneous occurring therapy naïve appendicular OSAs were analysed for the presence of ALT activity. Additionally, five already known canine sarcoma tissue samples with ALT activity as well as human ALT cell lines U2OS from osteosarcoma and YT-BO from astrocytoma served as positive controls. 14,35 Two human cell lines with telomerase activity served as negative controls.
These cell lines included SW480 derived from a colon adenocarcinoma and the cell line T98G (CRL-1690) derived from a glioblastoma multiforme.
Origin and handling of these cell lines was recently described. 14 Cell line validation statement: All cell lines except YT-BO are validated ATCC cell lines. Additionally, in all cell lines except U2OS cell line authentication testing was performed.

| DNA extraction and quantification
DNA from five to ten 10 μm thick sections of FFPE archived tumour tissue samples was extracted using nexttec 1-Step DNA Isolation Kit for Tissue and Cells (nexttec Biotechnology GmbH, Hilgertshausen-Tandern, Germany). After dewaxing with xylene and cell lysis overnight, DNA was extracted. 14 DNA isolation from human control cell lines was performed with Quick C-Circle Preparation (QCP) protocol. 19 A volume of 2 μL corresponding to 1/50 of nexttec and 1/25 of QCP extracted DNA were quantified with Qubit fluorometer (Invitrogen, Carlsbad, CA) using iQuant BR and HS dsDNA Quantification Kits (GeneCopoeia, Rockville, MD, USA) for detection of dsDNA between 0.2 and 1000 ng. Therefore, a minimum of 5 ng or 0.1 ng/μL extracted DNA could be measured. DNA samples were stored at −80 C.

| Radiolabel C-and G-circle assay and telomeric content
The presence of ALT activity was assessed using the radiolabel CCA method. 19 In detail, in a first step, extracted DNA was diluted to 32 ng/μL (dilution 1). For this, 2 μL of pure DNA extract were diluted in an appropriated amount of 10 mM Tris HCl pH 7.6 buffer for FFPE tissue samples or QCP buffer for human control cell lines.
QCP sample dilution 1 was performed with QCP buffer to standardize total amount of QCP buffer used in the CCA reaction for all samples. This process is necessary to minimize inhibitory effects of QCP lysis buffer. 19  units (AU) after background correction and normalization to alusignal as described. 19 All results were divided by 3 500 000 to make VSI numbers easier to handle. Then VSI of global background control obtained from samples with omitted DNA was subtracted. This was followed by correction for sample-specific background due to non-polymerase products. In detail, for each DNA sample the global background corrected VSI value from CCA reaction without poly- indicate ALT activity and AU less than 2-fold above background are considered as ALT-negative, because they are indistinguishable from background noise. 19 Levels that are between two-to 5-fold the background levels need to be carefully assessed with the appropriate controls.
Total telomeric content (TC) of all samples was calculated by using the so-called Telo/Alu method by dot blot analysis with radiolabeled gtel oligonucleotide. 36,37 In brief, telomeric dot blot VSI of phi-CCA reactions were normalized to corresponding values of Alu element repeats obtained from rehybridization experiment after stripping the membrane and after background correction as described above. 3 | RESULTS

| C-Circles detected in canine osteosarcomas above ALT threshold
Archived FFPE tumour tissue samples from 50 canine osteosarcoma patients were assessed for ALT activity using the radiolabel C-circle assay (CCA) ( Table 1). The CCA detected elevated CC levels above the published 5-fold background threshold in 10 (20%) cases ( Figure 1). Cases were grouped as CCA+ or CCA-according positive or negative CCA results (Table 1).
Input DNA content for the CCA ranged between 1 and 32 ng in 42 samples. In 8 of 50 samples, only low input DNA with less than 1 ng was available for the CCA. Four of the ten CCA+ cases were identified within these eight samples ( Figure 1A). Input DNA content did not correlate with the AU level, ruling out a dependency between these two parameters (Spearman r − 0.041,    Figure 1B).

| G-circles and telomeric content in canine osteosarcomas
In humans G-circles (GCs) are considered to be ALT specific although about 100-fold less abundant compared to CCs. 15

| ALT defined by CCA-status and associations with clinical characteristics
Demographic patient characteristics of the 50 dogs included into the study were grouped as ALT+ and ALT− according the CCA-status (Table 1).
Mean age of all 50 patients at the time of diagnosis was 8 years.
There was no difference in mean age at diagnosis between ALT+ and  Table 2). CCA− status remained a significant prognostic factor besides the factor "oncologic therapy" in the Cox proportional hazard model (P = .002) ( Table 3). None of the historical prognostic factors including "metastasis at diagnosis" had a significant influence on survival in our study ( Table 3). The latter finding might be explained by the fact that the few dogs with "metastasis at diagnosis" received aggressive treatment including salvage therapy.

| DISCUSSION
In this study we demonstrated by detection of partially singlestranded telomeric CCs that the telomerase-independent telomere maintenance mechanism ALT is active in 20% of canine appendicular osteosarcomas.
The detection of ALT specific CCs by the CCA has developed to a front-line ALT assay because it is a highly specific and sensitive assay. 16,19,20 An update of the original radiolabel CCA protocol as recently been published with slight modifications and was used in the present study. 15,19 In our previous study of 64 various canine sarcomas we already demonstrated that ALT+ human control samples showed comparable CC levels as canine samples and that the published human ALT thresholds were applicable for canine tumour specimens. 14,15,20 This comparability of ALT activity detection between the species was again confirmed in our present study. Samples with CC levels above the 5-fold background threshold were scored positive as recommended. 16,20 We further looked at the 2-fold background level because this level is described as suspicious for ALT activity. 19 From  DNA. Factors that are described to affect linearity are high amounts of genomic DNA (> 64 ng) that competes for the polymerase but not low amounts of DNA. 19 False-negative ALT activity in four other samples with less than 1 ng DNA applied for CCA could be excluded as comparable alu dot blot signals to the four CCA+ cases were detected.
Alu signals report the DNA input applied for the CCA. However, as ALT activity differs 10-fold within a range of 1. 31-12.74 AU between the identified ALT+ cases, tumours with weak ALT activity signals and less DNA applied for CCA might not be detectable as ALT cases.
G-circles (GCs) are partially single-stranded telomeric DNA circles that have as C-circles the ability to self-prime RCA of G-strand templates and are also considered as ALT specific. 15,16 Because GCs are about 100-fold less abundant than CCs they have hardly been investigated and are not used as a marker for ALT detection in humans. 16  Our finding is however consistent with the human literature, where telomere length alone is also not suitable for ALT detection. 20,42 Patient characteristics of our canine study cohort were consistent with a recent report on demographic characteristics and phylogenetic distribution of 744 dogs with appendicular osteosarcomas and other previous studies. 31,43,44 In the group of CCA+ patients, male dogs were overrepresented with a significant association between CCA− status and male gender. Several clinical studies in canine osteosarcomas report a slight overrepresentation of male dogs independent from any knowledge about TMM status. 30,45 No association of ALT phenotype with male gender has been reported in human osteosarcomas.
Our finding needs further investigations in a larger patient cohort.
Mean age at diagnosis was not significantly different between CCA+ and CCA− dogs. This is in concordance with some human reports. 27 Other human studies observed that ALT+ osteosarcoma patients were younger than ALT− patients and no ALT activity was found in patients older than 45 years. 26 In our cohort, no dog with an CCA+ osteosarcoma was older than 11 years. It can be speculated, that ALT phenotype occurs less frequently in senior patients in both species. In canine osteosarcoma, age-distribution is influenced by phylogenetic breed cluster. 31 This fact together with our observation that 9/10 dogs affected from CCA+ osteosarcomas were purebred dogs, suggests a contribution of genetic and possible inherited epigenetic factors in ALT activation. 17 The detection of an ALT phenotype in 20% of our study population is consistent with the findings of our initial study in 64 various sarcomas, in which ALT was detected in 25% of four osteosarcoma samples. 14 It is also in agreement with an earlier study that investigated telomerase activity (TA) in 67 canine osteosarcomas. 10  and this question is interesting for future investigations. The majority of our canine osteosarcomas studied most likely used TA and some might lack both TMMs altogether as described in previous studies. 10,14 Such tumours with lack of TMMs may represent a distinct class of cancer with a more favourable outcome. 9,27 In some human sarcoma subtypes ALT phenotype is linked to aggressive biologic behaviour and a negative prognosis. 23,24,26 In a recent meta-analysis in soft tissue sarcomas, the risk of mortality was higher for patients with ALT+ tumours. 9,49 In contrast, for human osteosarcomas the influence of ALT activity on survival is less clear. 26,27,29 The absence of any TMM has been reported as favourable prognostic factor 27 and the presence of TA as negative factor. 29 In both studies ALT status itself was not analysed as prognostic factor. In the latter study, the TA + group included only 14 tumours and 43% of them had additional ALT activity. It can be speculated that the ALT activity in nearly half of the TA+ osteosarcomas influenced their inferior outcome. Interestingly, only one human study investigated primarily the ALT status as prognostic factor for patient survival. 26  In summary, limitations of this study were a small heterogeneous study cohort, the retrospective survival analysis without standardized follow-ups and missing progression free survival times and the availability of only archived FFPE tissue specimens. Therefore, only gold standard and robust radiolabel CCA was performed for ALT detection.
No further ALT characteristics or evaluation of telomerase activity that can only be determined in fresh-frozen tissue could be analysed.

| CONCLUSIONS
Comparable to human osteosarcomas, a significant number of canine OSA use ALT as TMM. This high prevalence supports the notion of canine osteosarcomas as interesting model for comparative TMM research, especially as the dog is already an accepted spontaneous model for human osteosarcomas. In canine, the CCA− status might serve as a prognostic marker, but further research is needed to further test this hypothesis. Furthermore, canine studies may help to foster the assumed role of ALT to become an attractive tool for diagnosis and a target for therapeutic interventions in human cancer.