Alternative lengthening of telomeres in cancer confers a vulnerability to reactivation of p53 function

Abstract

A subset of cancers across multiple histologies with predominantly poor outcomes use the alternative lengthening of telomeres (ALT) mechanism to maintain telomere length, which can be identified with robust biomarkers. ALT has been reported to be prevalent in high-risk neuroblastoma and certain sarcomas, and ALT cancers are a major clinical challenge that lack targeted therapeutic approaches. Here, we found ALT in a variety of pediatric and adult cancer histologies, including carcinomas. Patient-derived ALT cancer cell lines from neuroblastomas, sarcomas, and carcinomas were hypersensitive to the p53 reactivator eprenetapopt (APR-246) relative to telomerase-positive models. Constitutive telomere damage signaling in ALT cells activated ATM kinase to phosphorylate p53, which resulted in selective ALT sensitivity to APR-246. Treatment with APR-246 combined with irinotecan achieved complete responses in mice xenografted with ALT neuroblastoma, rhabdomyosarcoma, and breast cancer and delayed tumor growth in ALT colon cancer xenografts, while the combination had limited efficacy in telomerase+ tumor models. A large number of adult and pediatric cancers present with the ALT phenotype, which confers a uniquely high sensitivity to reactivation of p53. These data support clinical evaluation of a combinatorial approach using APR-246 and irinotecan in ALT cancer patients.

Introduction:

Cancer cells overcome replicative senescence due to telomere shortening by activating telomerase or the alternative lengthening of telomeres (ALT) mechanism (1–3). ALT is observed in 5-10% of cancers with a higher prevalence in sarcomas and nervous system tumors (2,4), including pancreatic neuroectodermal tumors (4,5). ALT cancers often respond poorly to treatment (6–9). ALT employs homology-directed recombination-dependent replication to extend telomeres (10); however, the exact mechanism remains unknown. Therapeutic strategies have been proposed to exploit exclusive dependency of some tumors on ALT (11,12), but none have yet translated into the clinic.

The tumor suppressor p53 has a central role in response to cellular stress due to DNA damage, oncogene activation, hypoxia, and telomere shortening (13). In response to DNA damage, p53 can activate DNA damage repair, cell cycle arrest, and apoptosis (14). TP53 is the most mutated gene in cancer (15). The majority of ALT cell lines manifest dysfunction in the p53 pathway, most often by TP53 mutation or p53 degradation via viral oncoproteins (16). ATRX inactivating mutations are common in ALT cancers, but ATRX knockout did not activate ALT, whereas knock-out of both ATRX and TP53 induced ALT cancers in zebrafish (17). Finally, telomere dysfunction and telomere replication stress drive the ALT phenotype but lead to growth arrest and/or apoptosis in the presence of functional p53 (7,18,19). Although p53 pathway alterations are often observed in ALT cells (16), it is unlikely that p53 inactivation alone results in ALT, as the majority of TP53 mutant cancers are telomerase positive (TA+).

Mutations in TP53 are most commonly a single amino acid substitution that leads to p53 inactivation by disrupting p53 DNA binding interactions, or by causing conformational change in the protein, resulting in lower stability (20,21). Several small molecules are being explored as potential drugs to restore wild type p53 function in cancer. The clinical stage small molecule eprenetapopt (APR-246) (22) has been shown to induce cell death in cancer cells by restoring transcriptional activation and active conformation of non-functional p53 proteins (23,24). APR-246 is converted to a reactive electrophile methylene quinuclidinone (MQ), which covalently binds to the p53 core domain and stabilizes p53 and its complexes with DNA, leading to restoration of non-functional p53 proteins (24,25). High-resolution structural data demonstrated that the C124, C229, and C277 residues contribute in multiple ways to MQ-mediated stabilization of wild-type and mutant p53 bound to their DNA response elements. In addition to reactivation of p53, APR-246 is known to inhibit thioredoxin and depletes cellular glutathione, which leads to an increase in reactive oxygen species and presumably contributes to cytotoxicity in cancer cells (26–28). Although APR-246 has shown promising pre-clinical and clinical activity in several cancers (22,29–31), biomarkers that can identify patients with increased probability of response to APR-246 are needed to facilitate clinical development.

We recently demonstrated that constitutive ATM kinase activation (driven by telomere dysfunction) along with loss of p53 function induces chemotherapy resistance in ALT neuroblastomas (7). Here we expand those observations to a variety of adult and pediatric cancer histologies. As ATM phosphorylates and activates p53 (32), we hypothesized that restoration of p53 function using the p53 reactivator APR-246, especially in combination with DNA damaging agents, would be highly active against ALT cancers.

Materials and Methods

Drugs and Chemicals:

Topotecan hydrochloride hydrate, 7-ethyl-10 hydroxycamptothecin (SN-38), and puromycin dihydrochloride were obtained from Sigma-Aldrich, KU60019 from SelleckChem 4-Hydroperoxycyclophosphamide from Chemos GmbH & Co, irinotecan (IRN) hydrochloride injection from Sagent Pharmaceuticals, and APR-246 was provided by Aprea Therapeutics.

Patient-Derived Cell Lines (PDCLs):

All neuroblastoma and rhabdomyosarcoma cell-lines were obtained from the ALSF/Children’s Oncology Group (COG) Childhood Cancer Repository (www.CCcells.org), TX- series cell lines were established in our laboratory from patient samples obtained by the South Plains Oncology Consortium (www.SPONC.org) and all the remaining cell lines were obtained from American Type Culture Collection (ATCC). Source of all cell lines is indicated in supplementary table S1. All the cell lines were cultured in antibiotic-free Iscove’s Modified Dulbecco’s Medium, supplemented with 20% fetal bovine serum (FBS) (GIBCO), 1 X ITS and 4 mM L-glutamine. All cell lines were short tandem repeat (STR) assay verified and mycoplasma free at the time of experimentation.

Patient-Derived and Cell Line-Derived Xenografts (PDXs/CDXs):

Athymic nu/nu mice were injected subcutaneously at 6-8 weeks old with 10-15 million human cancer cells from cell cultures suspended in Matrigel to create a cell line-derived xenograft (CDX) or nu/nu mice were injected with human cancer cells from patients to create a patient-derived xenograft (PDX). Mouse weight and tumor volumes were measured 1-2 times weekly. Tumor volumes were calculated by 0.5*height*width*length as previously described (33). Mice were randomized into treatment groups at tumor volumes of 150 to 300 mm³ and sacrificed at endpoint which was ≥1500 mm³. The event-free survival (EFS) was defined as the time between randomization and endpoint, as previously described (34). All models were STR profiled and human vs mouse DNA was evaluated using qPCR at time of experiments as previously described (7). Details on xenografts used in this study are found in supplementary table S3.

Cytotoxicity Assay:

The DIMSCAN cytotoxicity assay was used to assess drug response in vitro, as previously described (35). Cytotoxic response to irinotecan (as SN-38), cyclophosphamide (4-HC), and doxorubicin was determined 96 hours post-treatment and assessed using in vitro concentration ranges with the highest being clinically achievable peak concentrations. To test the activity of APR-246, patient-derived cell lines (PDCLs) were treated for 96 hours and represented by their IC50 value (concentration cytoxic or inhibitory for 50% of live cells relative to untreated controls, calculated as previously described (35)). For activity of APR-246 + DNA damaging agents, cells were pre-treated for 72 hours with the DNA damaging agent before addition of APR-246. For some experiments the ATM inhibitor (KU60019) was given 24 hours prior to APR-246.

C-circle Assay:

C-circle assay was performed as previously described (6).

Immunofluorescence (IF) and Fluorescence in Situ Hybridization (IF-FISH):

IF-FISH was performed as previously described (7). Further details can be found in supplementary materials and methods.

Constitutive Knockdown and Overexpression Vectors:

Lentiviral plasmids (pLKO.1) containing p53 (TRCN0000003657) shRNA, ATM (TRCN0000010299) shRNA or eGFP shRNA (GE Healthcare Dharmacon, Inc, Lafayette, CO) were packaged using MISSION lentiviral packaging mix (Sigma, St Louis, MO) in HEK293FT packaging cells. PDCLs were infected with viral media for 72 hours followed by selection using 1.5 μg/mL of puromycin.

Saos2 cells were transiently transfected with 10 μg of plasmid DNA for each p53 variant (wild-type, R175H, R213*, R275H, R306*) or an empty vector control using FuGene HD Transfection Reagent (Promega, Madison, WI). After 48 hours, transfected Saos2 cells were collected for immunoblotting or seeded for DIMSCAN cytotoxic assay.

Retroviral vectors pLPC TRF2 deltaB deltaM, pLPC-MYC, pLPC-MYC-FokI^D450A-TRF1 and pLPC-MYC-FokI^WT-TRF1 were packaged using AmphoPack-293 cells. SK-N-BE(2) PDCL was infected with viral media for 72 hours followed by selection for >14 days using 1.5 μg/mL of puromycin.

In Vivo Drug Testing:

All in vivo experimentation was approved by the TTUHSC Institutional Animal Care and Use Committee (IACUC). We used 9 cell line-derived xenografts (CDXs: CHLA-90m, SK-N-FIm, SK-N-BE(2)m, Rh28m, Rh30m, Rh18m, TX-CC-199hm, TX-BR-100m, and MDA-MB-231m) and 3 patient-derived xenografts (PDXs: COG-N-669x, COG-N-519x, and TX-CC-286x). All CDX and PDX models were STR profiled to match the original patient and established in NOD scid gamma (NSG) mice before passage into athymic nu/nu mice for drug studies. Cells were strained from nude mice and injected subcutaneously with 10-15 million viable cells prepared in 200 μL of RPMI-1640 and Matrigel (Corning). Mice were randomized into control and treatment groups when tumors reached 150-300 mm³.

In vivo dosing of irinotecan (IRN) was designed to mimic clinical dosing (36); 20 mg/ml of irinotecan hydrochloride was diluted 10.67x in 0.9% saline; APR-246 was dissolved in 0.9% saline. In a 21-day cycle, on days 1-5 mice were treated with a tail vein injection at 7.5 mg/kg of IRN (~100 μL), followed by an intraperitoneal (IP) injection of APR-246 at 250 mg/kg twice daily 6 hours apart. APR-246 was administered on days 1-7 in a 21-day cycle. A total of 3 cycles were given for all groups.

In vivo responses to APR-246 and irinotecan were categorized based on the National Cancer Institute Pediatric Preclinical Testing Program (PPTP) classification system and can be seen in Supplementary Table S5 (37). Responses for each individual mouse for a given model were put into five categories: complete response (CR), maintained complete response (MCR), partial response (PR), stable disease (SD), and progressive disease (PD). A complete response (CR) is considered an unmeasurable tumor (<0.1 mm³) at any point in the experiment following treatment. A maintained complete response (MCR) is a CR that is maintained through the end of the experiment. A partial response (PR) is when a tumor reduces by >50% at least once during the experiment but is still a measurable mass. Stable disease (SD) is defined as a <50% reduction in the initial tumor volume with the tumor increasing no greater than 25% by the end of the experiment. A mouse is considered to have progressive disease (PD) when the tumor has a <50% reduction in tumor volume and is >25% at the end of the experiment. For each model, an event-free survival of the tumor relative to the control (EFS T/C) was calculated for each treatment group. The EFS T/C is calculated as the median survival of the treatment group mice divided by the median survival of the control. For those mice who do not have a median EFS that made it to the end of the experiment, the last recorded date is taken to calculate the median survival divided by that of the control group to determine the EFS T/C. An EFS T/C that is >2 and has a log-rank test that is statistically significant (P<0.05) is considered to be highly active while an EFS T/C < 2 has low activity.

Statistical analysis:

Comparison of two sample sets was done using Mann-Whitney U test. Comparison of biological or experimental replicates was done by unpaired two-tailed t test. Dose-response curves from cytotoxic assay were assessed using two-way analysis of variance. Combination Index (38) and IC50 concentration (35) were calculated as described previously. Survival analysis for xenograft studies was performed by the Kaplan-Meier method, as assessed using a log-rank test. All statistical analysis was performed using GraphPad Prism v7.0 and was considered statistically significant if P ≤ 0.05.

Data availability:

The data generated in this study are available within the article and its supplementary data files. PDCLs and PDXs are available from the COG/ALSF Childhood Cancer Repository under MTA (www.CCcells.org), from the corresponding author (CPR), or from ATCC. APR-246 can be requested from Aprea Therapeutics under MTA. Requests for other materials should be submitted to CPR.

See Supplementary Materials and Methods for additional details.

Results:

Patient-derived ALT cell lines manifest p53 dysfunction, constitutive double strand break (DSB) signaling, and resistance to DNA damaging agents

Most ALT cell lines used in early studies were derived from transformed fibroblasts with viral protein p53 inactivation (16). To assess the frequency of TP53 alterations or inactivation of the p53 pathway in patient-derived ALT cancer cells, we assembled a panel of 16 ALT PDCLs established from 14 patients (Supplementary Table S1) across 6 histologies [5 neuroblastoma, 3 rhabdomyosarcoma, 4 osteosarcoma, 2 colorectal adenocarcinoma, 1 lung adenocarcinoma, and 1 triple negative breast carcinoma (TNBC)]. All 16 ALT cell lines were positive for C-circles and 15/16 C-circle positive cell lines had low TERT mRNA expression (Fig. 1A and Supplementary Table S1). ALT associated PML bodies (APBs) were observed in 14 of 16 ALT PDCLs (Fig. 1A and Supplementary Table S1). Two colorectal ALT PDCLs (TX-CC-199h and TX-CC-208), established from the same patient, lacked APBs, which is in line with previous reports showing lack of APBs in some ALT cell lines (39).

Fig. 1. ALT PDCLs manifest dysfunctional p53 pathway, constitutive ATM kinase activation, and resistance to DNA damaging chemotherapy.

(A) Figure tracks ATRX mutations, TP53 mutations, mutations in p53 pathway related genes, telomeric C-circles, ALT associated PML-bodies (APBs), TERT mRNA expression, ATRX protein expression and p53 functionality in 16 ALT and 25 telomerase positive (TA+) cell lines across 6 histology (NB: neuroblastoma; OS: osteosarcoma; RMS: rhabdomyosarcoma; Colo: colorectal cancer; LU: lung cancer; TNBC: triple-negative breast cancer; Supplementary Table S1). Functionality of p53 is defined as high if there is >2-fold induction in p53 and p21 protein levels in cells following treatment with irradiation (10 grays) (B) Top panel shows representative immunofluorescence staining images for p-ATM (S1981) (green) in 3 ALT and 3 TA+ cell lines. Nuclei stained with DAPI (blue). Bottom panel displays p-ATM (S1981) mean fluorescence intensity (MFI) in 10 ALT PDCLs: versus 7 TA+ comparators across 6 histologies with irradiated (IR) SK-N-BE(2) as a positive control (cell line details are in Supplementary Table S1). A minimum of 100 cells for each PDCL were analyzed. (C) Comparisons of survival fraction (DIMSCAN cytotoxicity assay) for the ALT and TA+ PDCLs in A. Cell lines were treated with clinically achievable doses of DNA damaging agents topotecan, SN-38, and 4-hydroperoxycyclophosphamide. Wilcoxon-rank sum test was used for statistical analysis in B and C.

Whole exome sequencing (WES) data was assembled for the ALT PDCLs either from previous studies or acquired as a part of this study (Supplementary Table S1); source of WES data is provided in Supplementary Table S1 (6,16,40,41). ALT is commonly associated with ATRX loss-of-function genomic alterations (16,42) and expression of full length ATRX protein in ATRX altered ALT cells can suppress ALT activity (43). Here, we assessed ATRX genomic alterations and ATRX protein expression in sixteen ALT cell lines, five ALT PDCLs had a deletion in ATRX (Fig. 1A, Supplementary Table S1 and Supplementary Table S2) and lacked ATRX protein (Fig. 1A, and Supplementary Fig. S1). Two ATRX wild-type ALT PDCLs expressed very low levels of full length ATRX protein, while nine ALT wild-type ATRX PDCLs expressed high levels of ATRX protein; 23 of 25 randomly chosen TERT expressing PDCLs from the same histologies as the ALT PDCLs (Fig. 1A, and Supplementary Table S1; n = 25) expressed full length ATRX. Therefore, ALT activation can occur with and without ATRX loss, which is consistent with previous studies (5,6).

TP53 genomic alterations were identified in 11/16 ALT cell lines (8 missense mutations, 2 structural variants and 1 frame-shift mutation; Fig. 1A and Supplementary Table S1). Five ALT cell lines lacked any genomic alteration in TP53 or p53 pathway associated genes (Fig. 1A, and Supplementary Table S1), TP53 status for all cell lines was validated by Sanger sequencing. As loss of p53 function can occur in some ALT cells without a mutation, we assessed p53 functionality for all of the ALT and TA+ cell lines, as described previously (7). All the ALT cell lines, including TP53 wild-type ALT cell lines, were either unable to induce or showed a minimal increase (< 2 fold) in p53 and p21 protein expression levels upon irradiation, whereas 8 of 25 TA+ cell lines were functional for p53 (> 2 fold increase in p53 and p21 upon irradiation; Fig. 1A, Supplementary Fig. S2 and Supplementary Table S1; Fisher-exact test: P < 0.05).

We have shown that ALT neuroblastoma cells manifest constitutive DNA damage signaling at telomeres via activation of ataxia-telangiectasia mutated (ATM) kinase causing chemoresistance to DNA damaging agents (7). Here, we observed that ALT relative to telomer-ase-positive (TA+) PDCLs spanning 6 histologies showed higher levels of 53BP1 and phospho-ATM(S1981) foci, (Supplementary Fig. S3, A and B and Fig. 1B; Wilcoxon, P < 0.0001); 53BP1 foci co-localized with telomeres (Supplementary Fig. S3, C and D). The TA+ TNBC cell lines (MDA-MB-231 and TX-BR-162h) were the only TA+ cell lines that had high levels of 53BP1 and phospho-ATM(S1981) foci, comparable to that of ALT cells (Supplementary Fig. S3, A and B and Fig. 1B).

We compared cytotoxicity profiles of ALT (n=16) vs TA+ (n=25) PDCLs of 6 histologies treated with DNA damaging agents: topotecan (TOPO), SN-38 (irinotecan active metabolite), and 4-hydroperoxy cyclophosphamide (4-HC, cyclophosphamide active metabolite). Relative to TA+ PDCLs, ALT PDCLs were more resistant to DNA damaging agents (Fig. 1C; Wilcoxon, P<0.05).

The ALT phenotype (identified with the C-circle assay) has been reported previously to have a high prevalence (> 20%) in high-risk neuroblastoma and certain sarcomas (5,6). As we found ALT in carcinoma PDCLs, we assessed the prevalence of ALT in tumors from patients with breast, lung, and colon cancers. Extrapolating the number of ALT patients from the incidence data indicates ALT to be a large group of pediatric and adult patients identifiable with a specific biomarker that has a potentially targetable unique biology (Table 1).

Table 1.

Incidence of the Alternative Lengthening of Telomeres in Primary Solid Tumors.

Tumor Type	n	C-circle +	% Positive	Estimated Cases per year in USA	Estimated ALT Cases per year in USA
Pediatric
High Risk Neuroblastoma (6)	110	25	22.7%	250	56
Adult
Leiomyosarcoma (5)	49	38	77.6%	1346	1044
Triple-Negative Breast	20	2	10.0%	41250	4125
Non-Triple Negative Breast	158	3	1.9%	272000	5165
Colorectal Carcinomas	84	2	2.4%	104000	2476
Non-Small Cell Lung Cancer	27	2	7.4%	236,740	17536
Total Cases per Year					>30,000

Patient-derived ALT cell lines are sensitive to reactivation of p53 function using APR-246 in vitro

Due to activation of ATM kinase (which phosphorylates and activates p53) by the constitutive telomere DSB signaling in ALT cells, we hypothesized that reactivation of p53 function would be an effective therapeutic strategy in ALT relative to TA+ cancers. To test our hypothesis, we assessed cytotoxic profiles of ALT (n=16) versus TA+ (n=25) PDCLs in response to treatment with the clinical-stage p53 reactivator APR-246 in vitro (Supplementary Table S1). All the cell lines were cultured and tested for drug sensitivity at physiologic oxygen concentration (5% O₂). ALT relative to TA+ PDCLs were significantly more sensitive to APR-246 (P < 0.001), mean IC₅₀ ALT = 3.8 μM, TA+ = 10.5 μM and IC₉₀ ALT = 7.9 μM, TA+ = 19.2 μM). ; Fig. 2A and Supplementary Fig. S4). ALT relative to TA+ PDCLs were more sensitive to APR-246 in both TP53 wild-type and mutated cell lines (mean IC₅₀ in TP53 wild-type cell lines: ALT = 4.1 μM vs TA+ = 8.3 μM, P < 0.05; mean IC₅₀ in TP53 mutant cell lines: ALT = 2.5 μM vs TA+ = 13.2 μM, P < 0.05; Supplementary Fig. S5A) except the ALT osteosarcoma cell line Saos2 which has a null deletion in TP53 (Supplementary Fig. S5B). ALT PDCLs treated with APR-246 showed induction of p53 downstream transcriptional targets, CDKN1A, PMAIP1, BBC3 and BAX (Supplementary Fig. S6A). Consistent with induction of CDKN1A and PMAIP1 mRNA, p21 and NOXA protein levels were higher in cells treated with APR-246 (Supplementary Fig. S6, B and C).

Fig. 2. ALT PDCLs were sensitive to p53 reactivation using APR-246.

(A) Concentrations cytotoxic or inhibitory for 50% of cells (IC₅₀) in ALT (n=16) versus TA+ (n=25) PDCLs treated with APR-246 for 96 hours across 6 histologies; same PDCLs shown in Fig. 1A. (B) Top panel: immunoblotting for p53 and β-actin of TP53 missense mutated (CHLA-90) and wild-type PDCLs (COG-N-515) transduced with p53 or eGFP shRNA. Bottom panel: quantification of immunoblotting in B (means +/− SD from experimental triplicates). (C) Cytotoxicity assay curves in response to treatment with APR-246 in same cells as B. (D) Top panel: Immunoblotting for p53 and β-actin in cells transduced with plenti-empty vector, plenti-p53 WT, plenti-p53 R175H, plenti-p53 213*, plenti-p53 R273H, and plenti-p53 306* in an ALT TP53-null cell line (Saos2). Bottom panel: quantification of immunoblotting in D (means +/− SD from experimental triplicates). (E) Bar graph shows percent survival fractions in response to treatment with 5μM APR-246 in same cells as D. Cytotoxicity assay was performed using DIMSCAN. P-values compare cytotoxic response to APR-246 in cells transduced with same vectors as in D, relative to plenti-empty vector control. Sizes indicated on the immunoblot represent location of size markers but not the protein itself. Statistical significance was determined using two-tailed t-test in B and D, two-way analysis of variance (ANOVA) in E. *: P<0.05, **: P<0.01, ****: P<0.0001, ns: non-significant.

To evaluate the necessity of p53 protein expression for ALT sensitivity to APR-246, we knocked down p53 using lentiviral shRNA in 2 ALT neuroblastoma PDCLs, CHLA-90 (missense TP53 mutation) and COG-N-515 (wild-type TP53) (Fig. 2B). Both ALT PDCLs transduced with p53 shRNA were resistant to APR-246 relative to empty-vector controls (Fig. 2C; two-way analysis of variance (ANOVA): P < 0.001). Thus, either wild-type or mutant p53 protein is at least partly responsible for ALT sensitivity to APR-246, a result compatible with APR-246 inducing stabilization of both mutant and wild-type p53 (25,44). To assess the response of ALT cells to APR-246 with wild-type TP53 or common TP53 mutations (45), we overexpressed wild-type and 4 mutants of TP53 in the p53 null ALT cell line Saos2 (Fig. 2D). Two of the mutants were among the most common missense mutations in the TP53 DNA binding domain (R175H and R273H), one with a nonsense mutation causing a truncated protein with intact DNA binding domain (R306*) and partial p53 function (46), finally, a nonsense mutation that lacked full length DNA binding domain (R213*). Expression of p53 WT in Saos2 cells by itself did not induce mRNA expression of 3/4 p53 downstream transcriptional targets (CDKN1A, PMAIP1 and BAX), indicating that p53 expression by itself is not sufficient to induce p53 pathway in Saos2 cells (Supplementary Fig. S7A). However, Saos2 cells expressing p53 WT, p53 R175H, p53 R273H, or p53 R306* induced mRNA expression of 4 p53 transcriptional target (CDKN1A, PMAIP1, BBC3 and BAX) following treatment with APR-246 (Supplementary Fig. S7B) and were significantly more sensitive to APR-246 than the empty vector control (Fig. 2E). Cells expressing p53 R213* (lacks part of DNA binding domain) did not increase expression of CDKN1A, PMAIP1 and BBC3 mRNA following treatment with APR-246 (Supplementary Fig. S7B) and showed an APR-246 cytotoxic response similar to the empty vector control (Fig. 2E).

Thus, APR-246 is highly cytotoxic to ALT relative to TA+ cells due to reactivation of p53. ALT cell lines with a null TP53 or deletion of the TP53 DNA binding domain were less sensitive to APR-246.

Constitutive ATM kinase activation drives ALT sensitivity to APR-246.

It is well established that the ATM kinase activates p53 in response to DNA damage by phosphorylating at Ser15 (32,47). As ALT cells are associated with constitutive ATM activation, we assessed levels of phospho-p53 (p-p53, S15) in ALT and TA+ cell lines. ALT relative to TA+ cell lines had substantially higher levels of p-p53 (S15) (Fig. 3A and 3B). Additionally, we observed that p-ATM (S1981) and p-p53 (S15) immunofluorescence staining signal showed colocalization (Fig. 3C). Inhibition of ATM kinase activation using KU60019 abrogated p-p53 (S15) staining (Fig. 3D and 3E), indicating that ATM kinase activated p53 in ALT cells.

Fig. 3. Constitutive ATM kinase activation drives ALT sensitivity to APR-246.

A, Adjusted integrated density of p-p53(S15) for four ALT and three TAþ cell lines. B, Representative IF staining images for p-ATM(S1981) (green) and p-p53(S15) (red) in two ALT cell lines and one TAþ cell line. Nuclei were counterstained with DAPI (blue). C, Bar graphs representing Mander coefficient colocalization between p-ATM(S1981) and p-p53(S15) in same cells as B. D, Graph shows quantification of p-p53(S15) inALT cell line CHLA-90 +/− KU60019 (ATMinhibitor). E, Representative IF staining images for p-ATM(S1981) (green) and p-p53(S15) (red) in the ALT cell line CHLA-90 +/− KU60019. F, Immunoblotting for ATM and β-actin in ALT TP53-mutated (CHLA-90) and -WT (COG-N-515) PDCLs transduced with eGFP or ATM shRNA. G, DIMSCAN cytotoxicity assay curves in response to treatment with APR-246 in same cells as shown in F. H, Top, immunoblotting for p53, p21, NOXA, and β-actin in same PDCLs as in F, treated with APR-246. Bottom, quantification of immunoblotting in H and its replicates. I, DIMSCAN cytotoxicity assay curves in response to treatment with APR-246 +/− KU60019 in three ALT and three TAþ cell lines; fold change (FC) in IC50 for APR-246 with KU60019 is indicated on each graph. Sizes indicated on the immunoblot represent location of the size markers but not the protein itself. Statistical significance was determined using Wilcoxon rank-sum test in A and D, two-way ANOVA in G, two-tailed t test in H. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; NS, not significant. ATMi, ATM inhibitor; TP53-mut, TP53-mutated.

To determine if constitutive ATM kinase activation in ALT cells is required for the high sensitivity of ALT cell lines to reactivation of p53 function by APR-246, we assessed the effect of ATM shRNA knockdown on the cytotoxic and p53 transcriptional response to APR-246 in 2 ALT neuroblastoma cell lines (CHLA-90 and COG-N-515). Transduction of ATM shRNA induced resistance to APR-246 relative to control shRNA in both ALT cell lines (Fig. 3, F and G; two-way ANOVA: P < 0.001). Reduction in p21 and NOXA expression was observed in APR-246-treated ALT cells transduced with ATM shRNA relative to control shRNA, indicating reduced p53 activation in cells transduced with ATM shRNA (Fig. 3H; P < 0.05). Consistent with ATM knockdown, an ATM inhibitor (KU60019) antagonized the cytotoxicity of APR-246 in 4 ALT PDCLs (Fig. 3I and Supplementary Fig. S8; change in IC₅₀: P < 0.01) but showed only modest alteration in APR-246 cytotoxicity in TA+ PDCLs (Fig. 3I and Supplementary Fig. S8).

ALT cells have ATM-dependent telomere dysfunction induced foci (TIF) (7). To determine if vulnerability to p53 reactivation by APR-246 is associated with TIF, we induced ATM-dependent TIF in a TP53 mutant, p53 non-functional TA+ neuroblastoma PDCL (SK-N-BE(2); Supplementary Table S1) by overexpressing dominant-negative TRF2 (TRF2ΔBΔM; Fig. 4A) via transduction with the pLPC-nMyc-TRF2ΔBΔM retroviral vector or pLPC-nMyc (empty vector control), as described previously (7,48,49). Expression of TRF2ΔBΔM in SK-N-BE(2) induced a significant increase in phospho-ATM(S1981) (Fig. 4, A and B) and TIF relative to control (Fig. 4C; P < 0.05), that was abrogated when treated with an ATM inhibitor (Fig. 4C; P < 0.01), indicating an ATM-dependent TIF response. SK-N-BE(2) expressing TRF2ΔBΔM were hypersensitive to APR-246 cytotoxicity relative to control (two-way ANOVA: P < 0.001; Fig. 4D) and sensitivity was reversed with an ATM inhibitor to a level of cytotoxicity observed in empty vector controls (two-way ANOVA: P = ns; Fig. 4D). Previous work by another group revealed that overexpression of TRF2 can partially suppress ATM dependent TIF response in ALT cells (50). To assess if overexpression of TRF2 suppresses ATM activation and desensitizes ALT cells to APR-246, we transiently overexpressed TRF2 using pLPC-nMyc-TRF2 vector in TP53 mutant ALT cell line Rh28, which is hypersensitive to APR-246. Overexpression of TRF2, reduced the expression of p-ATM(S1981) (Supplementary Fig. S9, A to C) and induced resistance to APR-246 (two-way ANOVA: P < 0.001; Supplementary Fig. S9D).

Fig. 4. Induction of ATM dependent telomere dysfunction signaling in a telomerase positive cell line caused hypersensitization to APR-246.

(A) Immunoblotting for TRF2 and β-actin in TA+ cell line SK-N-BE(2) transduced with pLPC-nMyc or pLPC-nMyc-TRF2ΔBΔM. (B) Quantification for immunoblotting in A and its replicates. (C) Top panel: representative images of immunofluorescence detection of 53BP1 (red) in combination with fluorescent in situ hybridization to detect telomeres (green) in same cells as A +/− ATM inhibitor (KU60019). Bottom panel: bar graph represents mean TIF count in C and its replicates. A minimum of 50 cells were assessed for each replicate. (D) DIMSCAN cytotoxicity assay curves for same cells as A treated with APR-246 +/− ATMi. Sizes indicated on the immunoblot represent location of the size markers but not the protein itself. Statistical significance assessed using a two-tailed t-test in B and C; with two-way ANOVA in D. **: P<0.01, ns: not significant.

As an alternative approach to induce ATM-dependent TIF in TA+ cells, we overexpressed the nuclease domain of FOKI (nonspecific endonuclease) fused to the N-terminal domain of shelterin protein TRF1 (FokI^WT-TRF1) in SK-N-BE(2) using a retroviral vector, with cells transduced with nuclease-dead mutant (FokI^D450A) fused to TRF1 (FokI^D450A-TRF1) as a control, as described previously (51,52). Both the fusion proteins FokI^WT-TRF1and FokI^D450A-TRF1 localized to telomeres (Supplementary Fig. S10A). Cells expressing FokI^WT-TRF1 induced higher phospho-ATM (S1981), ATM-dependent TIFs, and cytotoxic sensitivity to APR-246 relative to cells expressing FokI^D450A-TRF1 or empty vector control (Supplementary Fig. S10, B and C; P < 0.01; Supplementary Fig S10, D and E; Supplementary Fig S10F, two-way ANOVA: P < 0.0001).

Thus, ATM knockdown or inhibition abrogated APR-246 activity in ALT cell lines. Induction of an ATM-dependent TIF response in a TA+ cell line caused hypersensitivity to APR-246.

APR-246 potentiated irinotecan activity in ALT cancer cell lines and xenografts.

Loss of p53 function is often associated with chemoresistance in cancer (53), and we observed that ALT PDCLs have loss of p53 function and are resistant to DNA damaging chemotherapy across multiple histologies (Fig. 1, A and C). To determine if exogenous DNA damage combined with APR-246 is more cytotoxic than APR-246 alone, we assessed the cytotoxic response to APR-246 with or without irinotecan (as SN-38, in vitro) pretreatment in 8 ALT versus 8 TA+ PDCLs across 6 histologies; ALT PDCLs used for this cytotoxicity assay were randomly selected for each histology, SN-38-resistant TA+ cell lines were chosen for comparison. Irinotecan was used due to its activity and clinical use across multiple histologies (54–57). ALT cell lines across all histologies uniformly showed an increased cytotoxic response to SN-38 + APR-246 compared to either SN-38 or APR-246 alone (Fig. 5A and Supplementary Fig. S11A), whereas TA+ cell lines showed a heterogeneous response to the combination (Fig. 5B and Supplementary Fig. S11B). Although APR-246 + SN-38 was synergistic in most PDCLs, ALT relative to TA+ PDCLs had a significantly higher synergistic response to the combination (Fig. 5C; combination index: P < 0.05). Caspase-3 and PARP cleavage was higher in ALT PDCLs treated with APR-246+SN-38 relative to single agents (Fig. 5, D and E), indicating that synergy was driven by apoptotic cell death.

Fig. 5. APR-246 reversed chemoresistance and sensitized ALT PDCLs to IRN (SN-38) in vitro.

A and B, DIMSCAN cytotoxicity assay curves in ALT (n = 4; A) and TA+ (n = 4; B) PDCLs treated with APR-246 + SN-38. C, Comparison of CI values for APR-246 + SN-38 at top two equimolar doses in ALT versus TA+ cell lines. D, Immunoblotting for PARP, cleaved–PARP, cleaved caspase-3, and β-actin in ALT cell lines (CHLA-90 and TX-BR-100) and a TA+ cell line [SK-N-BE (2)] treated with APR-246 + SN-38. E, Normalized quantification of cleaved PARP and cleaved caspase-3 in same cell lines as D. The bar graphs represent the means with SD in experimental triplicates. Sizes indicated on the immunoblot represent location of the size markers but not the protein itself. Statistical significance was assessed using two-way ANOVA in A and B, Wilcoxon rank-sum test in (C), and two-tailed t test in E. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; NS, not significant. TP53-mut, TP53-mutated.

We assessed the activity of APR-246 +/− irinotecan in vivo in 7 ALT (3 neuroblastoma, 2 rhabdomyosarcoma, 1 colorectal and 1 TNBC) and 5 TA+ (2 neuroblastoma, 1 rhabdomyosarcoma, 1 colorectal and 1 TNBC) patient-derived (PDX) and PDCL-derived murine xenograft models (Fig. 6; Supplementary Table S3). ALT xenografts had missense mutations in TP53 except for COG-N-669x which has wild-type TP53. Choice of xenograft models was based on ability to grow progressively in mice. A minimum of one TA+ xenograft model per histology were assessed, TA+ models with higher resistance to SN-38 were chosen as comparators.

Fig. 6. ALT relative to telomerase-positive (TA+) xenograft models are sensitive to irinotecan + APR-246 in vivo.

Tumor growth curves in (A) 4 ALT PDX/CDXs and (B) 4 TA+ PDXs/CDXs across 4 histologies treated with irinotecan or irinotecan + APR-246. Kaplan-Meier event-free survival (EFS) curves combined for 4 histologies in (C) 7 ALT PDX/CDXs [neuroblastoma (CHLA-90, SK-N-FIm, COG-N-669x), rhabdomyosarcoma (Rh28m, Rh30m), colorectal adenocarcinoma (TX-CC-199hm) and triple-negative breast carcinoma (TX-BR-100m)] and (D) 5 telomerase expressing PDXs/CDXs [neuroblastoma (SK-N-BE(2)m, COG-N-519x), rhabdomyosarcoma (Rh18m), colorectal adenocarcinoma (TX-CC-286x) and triple-negative breast carcinoma (MDA-MB-231m)]; PDCL and PDX characteristics are in Supplementary Table S1 and S3, respectively). Mice were treated for three 21-day cycles with APR-246 (250 mg/kg) on days 1-7; irinotecan (7.5mg/kg) on days 1-5, or both agents simultaneously. Dosing was initiated at tumor volumes between 150 and 300 mm³. EFS was defined as the time taken for tumor volume to reach 1500 mm³. Each experiment had 5 to 9 mice in control and treatment groups (details in Supplementary Table S5). Statistical significance for EFS was determined using the log-rank test (individual xenograft log-rank test data are in Supplementary Table S4).

APR-246 by itself showed low activity in 4/7 ALT and high activity in 1/7 ALT xenograft models, whereas only 1/5 TA+ xenograft models treated with APR-246 had an objective response (Supplementary Table S4). Across all histologies, APR-246 increased event-free survival (EFS) relative to controls in ALT (Fig. 6 A and C) but not in TA+ xenografts (Fig. 6 B and D) (log-rank test: P < 0.0001). A TA+ TNBC xenograft model (MDA-MB-231m), that had constitutive (pre-therapy) higher levels of double strand breaks and p-ATM (S1981) foci (Supplementary Fig 1B and Supplementary Fig. S1A), was the only TA+ xenograft that achieved objective responses in mice treated with APR-246 relative to control (Supplementary Table S4 and Supplementary Table S5). APR-246 significantly enhanced activity of irinotecan in all ALT xenografts (Fig. 6 A and C) but not in TA+ xenografts (Fig. 6 B and D; Supplementary Table S4 and Supplementary Table S5). In 3 ALT neuroblastomas (CHLA-90m, SK-N-FIm, and COG-N-669x), 2 ALT rhabdomyosarcomas (Rh28m and Rh30m), and 1 ALT TNBC (TX-BR-100m) xenograft models, nearly all mice (47/48) treated with irinotecan + APR-246 achieved either a complete response (CR) or maintained complete response (MCR) and had increased EFS relative to mice treated with either irinotecan or APR-246 alone (Supplementary Fig. S12A, S13, S14A, Supplementary Table S4 and Supplementary Table S5). The ALT colorectal cancer xenograft (TX-CC-199hm) had enhanced EFS and delayed tumor growth when treated with APR-246 + irinotecan compared to single agents (Supplementary Fig. S14B; Supplementary Table S4 and Supplementary Table S5). By contrast, the TA+ neuroblastoma, rhabdomyosarcoma, colorectal, and TNBC xenografts had intermediate or low responses to APR-246 + irinotecan, with no further enhancement in EFS compared to mice treated with either APR-246 or irinotecan (Supplementary Fig. S12B, S13, S14, Supplementary Table S4 and Supplementary Table S5). APR-246 + irinotecan was well tolerated by the mice (Supplementary Fig. S15). Histopathological examination of xenografts treated with APR-246 + irinotecan at day 7 of therapy showed reduced proliferative activity (Ki-67) and increased apoptosis by TUNEL relative to untreated or single-agent-treated tumors in ALT but not in a TA+ xenografts (Supplementary Fig. S16).

Discussion:

Loss of p53 function is thought to be required for the ALT mechanism, as p53 is known to induce cell cycle arrest and apoptosis in response to telomere dysfunction and telomere replication stress (7,18,19) and constitutive ATM activation (7,18,19) that is specific to ALT cells. Most ALT cells lack functional p53 and contain remarkably large numbers of telomere dysfunction induced foci (TIF) (7,50). Here we confirmed that ALT patient-derived cell lines (PDCLs) across multiple histologies (both adult and childhood cancers) are associated with extensive TIFs and constitutively activate ATM kinase and p53 non-functionality. Surprisingly, the cell line U2OS, which is thought to have p53 function (58), showed very minimal increase in p53 and p21 protein expression following irradiation, relative to positive control. Thus, in our hands U2OS shows, non-responsiveness of p53 pathway to irradiation induced DNA damage. The discrepant results observed in our work could possibly be due to clonal variation in the cell line.

Our results show that reactivation of p53 function using APR-246 is significantly more cytotoxic to ALT compared to TA+ cells and that an intact p53 DNA binding domain is partially required for ALT sensitivity to APR-246, indicating likely resistance to APR-246 in tumors that have loss of p53 expression by a deletion in TP53 or MDM2 amplification, such as osteosarcomas and liposarcomas, respectively (59,60). Nevertheless, as the majority of TP53 mutations in cancer occur as missense mutations in the DNA binding domain (45), a large proportion of ALT cancers are likely sensitive to APR-246.

Consistent with our previous observations in neuroblastoma (7), we show here that ALT cells are associated with constitutive ATM kinase activation, and telomere dysfunction across multiple histologies. Interestingly, not all 53BP1 foci (DNA double-strand break (DSB) marker) colocalized to telomeres, suggesting the possibility that some DSB signaling in ALT cells can emanate at other genomic loci. Nevertheless, our data indicate that ALT cells appear to tolerate ongoing genomic stress. Our results indicate that ATM activation found in ALT results in susceptibility to p53 reactivation using APR-246. Additionally, we showed that induction of ATM-dependent TIFs in a TP53 mutant telomerase-positive cell line conferred sensitivity to APR-246. However, as other forms of telomere replication stress have been observed in ALT cells (10,11), ATM kinase activation may not be the only mechanism for the high ALT sensitivity to p53 reactivation.

We have previously shown that loss of p53 function combines with ATM dependent telomere dysfunctional signaling to induce chemoresistance to DNA damaging agents in ALT neuroblastoma (7). Here we showed that ALT cell lines across multiple histologies are generally insensitive to DNA damaging agents across multiple histologies. We demonstrated that due to the high constitutive levels of activated ATM kinase found in all ALT cells, that ALT cell lines were especially sensitive to p53 reactivation with APR-246. We also observed a high degree of synergistic cytotoxicity in ALT cell lines for APR-246 + irinotecan (as SN-38 in vitro), while TA+ cell lines showed only modest to low synergy between APR-246 and SN-38. Thus, the ATM activation from constitutive telomere damage signaling in ALT cells resulted in high p53 activation by APR-246 and APR-246 combined with the DNA damage induced by irinotecan achieved a high degree of synergistic cytotoxicity in ALT cell lines. This proposed mechanism of action is illustrated in Fig. 7.

Fig. 7. Proposed mechanistic illustration of APR-246 activity in ALT cells.

Telomerase positive cells have proficient telomere end protection, which inhibits activation of DNA damage signaling via ATM kinase. ALT cells are associated with constitutive ATM activation at telomeres. Loss of p53 function in ALT tumor cells dramatically reduces ATM kinase contribution to apoptosis. When p53 function is restored using APR-246, ATM signaling in ALT cells induces apoptosis.

Consistent with the in vitro cytotoxicity data, APR-246 + irinotecan was highly active against ALT xenografts from neural tumors (neuroblastoma), sarcomas (rhabdomyosarcoma), and carcinomas (TNBC) achieving prolonged (> 100 days) event-free survival in most, whereas TA+ xenografts showed little to no benefit from combining APR-246 and SN-38. The TNBC TA+ PDCL-derived xenograft (MDA-MB-231m) had a higher number of innate DSBs, manifested ATM activation comparable to ALT PDCLs, and was more sensitive to APR-246 relative to other TA+ xenografts. Interestingly, as MDA-MB-231 cell line had very low telomere content relative to other telomerase positive cell lines, it is possible that these cells might have TIFs due to excessively shortened telomeres (50). As we showed that constitutive ATM signaling at telomeres can make cells susceptible to APR-246, it will be interesting to investigate if APR-246 has high activity in p53 non-functional TA+ cells with shorter telomeres in future studies. Irrespective of the location of DSB, our data in this study suggest that innate DSB signaling occurring in some TA+ cancers can increase APR-246 sensitivity. It is possible that the high sensitivity to APR-246 observed in ALT cells may also be due in part to other mechanisms such as GSH depletion, as elevated intracellular oxidative stress is associated with sensitivity to APR-246 (27,61).

One of the limitations of our study is the wide range of TP53 genomic alterations and histologies in our cell line panel, making it not feasible to compare APR-246 sensitivity by specific mutation between ALT and TA+ cell lines. Another limitation of our study is the lack of isogenic ALT and TA+ cell line, as the majority of known ALT and TA+ isogenic pairs are fibroblast transformed cell lines that lack p53 protein.

In summary, ALT mediated TMM is found in a variety of pediatric and adult cancer histologies and can be detected with robust biomarkers (DNA C-circles, ALT-associated PML bodies, or heterogeneous telomere foci) (4). ALT cancers commonly manifest resistance to current therapies and are in need of novel therapeutic options. Here we demonstrated that reactivation of p53 function with APR-246 is cytotoxic to ALT cells. APR-246 cytotoxicity was mediated by the constitutive high levels of ATM activation found in ALT cells that enhances p53 activity. APR-246 or APR-246 + irinotecan showed low activity against most telomerase+ cancer cell lines and xenografts. However, APR-246 + irinotecan was synergistically cytotoxic for most ALT cancer cell lines and achieved durable complete responses in ALT xenograft models across multiple pediatric and adult cancer histologies. These data support carrying out clinical trials to test APR-246 + irinotecan in patients with ALT cancers.

Significance:

This work demonstrates that constitutive activation of ATM in chemotherapy-refractory ALT cancer cells renders them hypersensitive to reactivation of p53 function by APR-246, indicating a potential strategy to overcome therapeutic resistance.