Induction of Telomere Dysfunction Prolongs Disease Control of Therapy-Resistant Melanoma

Telomerase, a ribonucleoprotein enzyme complex, is expressed in almost 90% of human tumors while most somatic tissues show no enzyme activity except in transiently proliferating stem-like cells. Therefore, telomerase remains a highly attractive target for the development of new anticancer agents. Telomerase counteracts telomere shortening by adding TTAGGG repeats at the end of chromosomes via its reverse transcriptase ribonucleoprotein complex, which is minimally composed of the catalytic protein subunit hTERT and functional RNA component hTR or hTERC.

Aberrant regulation of telomerase activity is one of the hallmarks that distinguish cancer cells from normal cells because advanced cancer cells can almost universally divide indefinitely. Owing to the maintenance of telomeres, this phenomenon is almost universally orchestrated by telomerase (1). Recent studies have revealed a high frequency of TERT promoter mutations, C250T and C228T in melanomas, surpassing other types of human cancers (2–10). TERT promoter mutations generate de novo consensus binding motifs for ETS transcription factors (11). Noncanonical NFκB signaling and ETS1/2 then cooperatively drive the activation of the C250T-mutant TERT promoter and increase TERT levels and enzyme activity (12). The de novo ETS motif is critical for mutant TERT promoter activity and facilitates heterotetramer binding of GABP to the mutant TERT promoter (13).

BRAF-mutated cutaneous melanomas inevitably acquire resistance to inhibitors targeting the MAPK pathway (MAPKi) that are active in BRAF-mutated tumors. Molecular mechanisms underlying both intrinsic and acquired resistance to mutant BRAF-targeted therapies are highly complex and variable (14–16). This warrants the identification of actionable therapies that are personalized and tailored to each therapy-resistant tumor's genetic composition and proteomic profiles.

Immune checkpoint blockade therapies have also emerged as first-line therapies for treating patients with unresectable or metastatic melanomas. For treatment-naïve patients enrolled in a randomized phase III study, the objective response rate was 19%, 43.7%, and 57.6%, respectively, for ipilimumab, nivolumab, and nivolumab plus ipilimumab (17). However, depending on the therapy arm, almost half of the patients progressed despite the initial success of immunotherapy. This suggests that combinational or additional drugs administered in sequence might potentially prolong clinical responses.

Taken together, there remains a large unmet and urgent need to identify novel therapies for treating patients with metastatic melanoma who have had disease progression after first-line therapies. Our previous study demonstrated that the induction of telomere dysfunction by telomerase-directed 6-thio-2′-deoxyguanosine (6-thio-dG) therapy significantly impairs viability and proliferation of telomerase-positive lung and colon cancer cells but not telomerase-negative normal cells (18). We reason that targeting telomerase might provide a novel therapeutic approach for treating patients with unresectable or metastatic melanoma whose disease has progressed. In this study, we show the broad efficacy of the telomerase-directed 6-thio-dG in multiple preclinical models of therapy-resistant human and mouse melanomas.

All clinical data and patient samples were collected following approval by the Massachusetts General Hospital Institutional Review Board (IRB) and the Hospital of the University of Pennsylvania IRB. In all cases informed consent was obtained from patients. All animal studies were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the NIH. Mice were maintained according to the guidelines issued by Institutional Animal Care and Use Committee (IACUC) at The Wistar Institute (Philadelphia, PA). Study designs were approved by IACUC at The Wistar Institute (Philadelphia, PA).

All normal skin epidermal melanocytes, keratinocytes, and human metastatic melanoma cell lines that were established at The Wistar Institute (Philadelphia, PA) and methods that were used for the routine authentication and mycoplasma test of cell lines have been documented in https://www.wistar.org/our-scientists/meenhard-herlyn. UACC-62 and UACC-903 cells were kind gifts from Dr. Marianne B. Powell (Stanford University, Stanford, CA). A375 cells were purchased from ATCC. LOX-IMVI cells were kindly provided by Dr. Lin Zhang (University of Pennsylvania, Philadelphia, PA). 499 and JB2 cells were kind gifts from Dr. Andy Minn (University of Pennsylvania, Philadelphia, PA). All resistant cell lines that acquired drug resistance to BRAF inhibitor or the combination of BRAF inhibitor plus MEK inhibitor were described previously (19, 20). All cell lines were maintained in RPMI1640 media (Mediatech, Inc.) supplemented with 10% FBS (Tissue Culture Biologicals) and cultured in a 37°C humidified incubator supplied with 5% CO₂.

The BRAF inhibitor PLX4720 was provided by Plexxikon Inc. The BRAF inhibitor dabrafenib, the apoptosis inducer staurosporine, and the pan-caspase inhibitor Z-VAD-FMK were purchased from LC Laboratories. The MEK inhibitor PD0325901 was purchased from Selleckchem. 6-thio-dG used for in vitro studies was purchased from Metkinen Chemistry Oy. 6-thio-dG used for in vivo studies was purchased from R I Chemical Inc. The telomerase inhibitor, BIBR 1532 was purchased from Cayman Chemical. The lipid modified 13-mer thio-phosphoramidate oligonucleotide is a telomerase template antagonist GRN163L (Imetelstat) as described previously (21).

A total of 1 × 10⁵ melanoma cells were harvested from cell culture and resuspended in culture medium and Matrigel at a 1:1 ratio. Cells were subcutaneously injected into mice, which were treated with indicated inhibitors once daily via intraperitoneal (IP) administration (6-thio-dG) or oral gavage (dabrafenib) when the tumor volume reached the range of 50–100 mm³. Mice were sacrificed at the end time point and solid tumors were collected. All animal experiments were performed in accordance with Wistar IACUC protocol 112330 in NOD.Dg-Prkdc scidlL2rg tm 1 Wjl/SzJ mice.

RNA purification was done using the AllPrep DNA/RNA Mini Kit (Qiagen) for 31 tumor biopsy specimens. The first batch of 17 RNA samples were ribo-zero treated and then subject to library preparation using Epicentre ScriptSeq Complete Gold kit. Quality check was done on the Bioanalyzer using the High Sensitivity DNA kit and quantification was carried out using KAPA Quantification kit. Samples were sequenced on Illumina NextSeq500 with the 2 × 75 bp high output in the Genomics Core Facility at The Wistar Institute (Philadelphia, PA). The second batch of 14 RNA samples purified from tumor biopsy specimens along and the third batch of 12 RNA samples purified from A375 and LOX-IMVI BR cells were sequenced at Broad Institute to achieve the high coverage of 50M pairs. Briefly, the Tru-Seq Non-Strand Specific RNA Sequencing, which includes plating, poly-A selection, and nonstrand-specific cDNA synthesis, library preparation, sequencing, and sample identification QC check (when Sample Qualification of a matching DNA sample is chosen). RNA-seq data of A375 and LOX-IMVI BR cells were deposited into Gene Expression Omnibus (accession number GSE99552).

Adherent cells were harvested with 0.05% Trypsin-EDTA, pooled with floating cells and then washed once with 1× DPBS. Cells were then pelleted and stained with PSVue 643 at 0.5 μmol/L and propidium iodide at 50 μg/mL diluted in TES buffer for 5 minutes in the dark. Cells were then immediately subjected to FACS analysis using a BD LSR II flow cytometer and at least 5,000 cells per sample were acquired.

Cells were seeded into 12-well tissue culture plates at a density of 500 cells/well as biological triplicates in drug-free medium. Medium was refreshed every 3 or 4 days for 14 days. Colonies were then stained overnight with methanol containing 0.05% crystal violet. After extensive washing with distilled H₂O, cells were air-dried and subjected to image acquisition using a Nikon D200 DLSR camera.

The source and analysis of gene expression microarray data, RNA-sequencing (RNA-seq) data, and reverse-phase protein array (RPPA) of melanoma cell lines or paired pre- and posttreatment tumor biopsies derived from patients with metastatic melanoma were previously described (19, 20).

The quantitation of telomerase enzyme activity was performed as described previously (22).

The hematoxylin and eosin (H&E) and IHC staining of tumor and liver specimens was performed as described previously (19, 20). Ki-67, AXL, and phospho-S6 antibodies were purchased from Cell Signaling Technology (catalog nos. are 9027, 8661, and 4858).

pMKO.1-GFP (https://www.addgene.org/10676/) and pMKO.1-shTERT (https://www.addgene.org/10688/) plasmids obtained from Addgene. For the retroviral production, the platinum A cell line (Cell Biolabs) was transfected with pMKO.1-GFP and pMKO.1-shTERT plasmids using the Calcium Phosphate method. Twenty-four hours after transfection, the fresh medium was added. Starting from day 2 after the initial transfection, viral supernatants were collected for two consecutive days. Viral supernatants were centrifuged at 1,200 rpm for 5 minutes and filtered using 0.45-μm filters. A375 cells were infected with viral supernatants supplemented with Polybrene at 8 μg/mL. Stably infected cells were established by either FACS sorting of GFP-positive cells or puromycin selection.

Unless otherwise indicated, data in the figures were presented as mean ± SEM for three biological or technical replicates. Significant differences between experimental conditions were determined using the two-tailed unpaired t test. For survival data, Kaplan–Meier survival curves were generated and their differences were examined with log-rank test. For tumor growth data, mixed effect models were used to determine the differences between treatment groups in tumor volume change at the end of experiment. A two-sided P value of less than 0.05 was considered statistically significant (*, P < 0.05; **, P < 0.005; ***, P < 0.0005).

We previously demonstrated that 6-thio-dG inhibited cell viability of the colon cancer cell line, HCT116 and the non–small cell lung cancer cell line, A549 (18). Next, we set out to test the efficacy of 6-thio-dG in the genetic subtype of BRAF-mutated melanoma. We treated the first cohort of 12 human metastatic melanoma cell lines harboring the BRAF^V600E mutation with 6-thio-dG for 9-12 days. As controls for 6-thio-dG, we also tested the analogue of 6-thio-dG, 6-thioguanine and a well-known telomerase inhibitor, BIBR 1532 (23). Interestingly, we observed a marked antiproliferative effect of 6-thio-dG in BRAF-mutated melanoma cell lines, which is superior compared with that of BIBR 1532 (Fig. 1A). In contrast, we showed that 6-thio-dG did not impair the cell viability of normal human skin cells, including keratinocytes (FK170), melanocytes (FOMG005), and fibroblasts (GM21), whereas the apoptosis inducer Staurosporine significantly induced apoptosis and cell death by day 4 (Fig. 1B; ref. 24). Our findings are consistent with our previous study showing that human normal fibroblasts and colonic epithelial cells were not as sensitive to 6-thio-dG as telomerase–positive cancer cells (18). Taken together, our results suggest telomerase-positive BRAF-mutated melanoma cells are sensitive to 6-thio-dG.

The investigation of BRAF-mutated melanoma cell lines allowed us to compare the efficacy of 6-thio-dG with that of a BRAF inhibitor, PLX4720. We treated the second cohort of 16 BRAF-mutant melanoma cell lines with 6-thio-dG and PLX4270 for 9–12 days, respectively. There were three melanoma cell lines that overlapped in both cohorts, including A375, SK-MEL-28, and WM793 (Figs. 1A and 2A–D). We demonstrated that 6-thio-dG also substantially inhibited cell viability of BRAF-mutated melanoma cell lines in the second cohort and the efficacy of 6-thio-dG was comparable with and in some cases even superior to that of PLX4720 (Fig. 2A–D; Supplementary Table S1).

TERT promoter mutations are known for 9 and 16 BRAF-mutated human melanoma cell lines included in the first (Fig. 1A) and second cohort (Fig. 2A–D), respectively. WM278, WM35, and WM793 melanoma cell lines harbor wild-type TERT promoter and they responded well to 6-thio-dG (Fig. 2A–D). Therefore, we reasoned that the status of TERT promoter mutation is unlikely to be associated with responses of BRAF-mutated melanoma cell lines to 6-thio-dG.

Treatment of 12 BRAF-mutated melanoma cell lines with 6-thio-dG over a shorter period of 5 days significantly induced apoptosis and cell death in 9 of them (Supplementary Fig. S1A). A375 was the one with the highest sensitivity to 6-thio-dG (Supplementary Fig. S1A). Using A375 as a representative cell line, we showed that the addition of a pan-caspase inhibitor (Z-VAD-FMK) significantly decreased the percentage of apoptotic and dead cells treated with 6-thio-dG for 72 hours (Fig. 2E). In a time-course experiment, we further demonstrated that the induction of apoptosis and cell death in A375 cells treated with 6-thio-dG at 2.5 and 5 μmol/L could occur as early as 48 hours posttreatment (Fig. 2F).

The induction of apoptosis and cell death prompted us to also investigate cells that had survived 6-thio-dG. The senescence-associated β-gal (SA-β-gal) staining of 4 BRAF-mutated melanoma cell lines suggested that long-term treatment with 6-thio-dG triggered induction of cellular senescence in cells that survived the initial killing by 6-thio-dG (Supplementary Fig. S1B).

Finally, we established xenografts of 1205Lu and A375 melanoma cell lines and showed that 6-thio-dG significantly impaired the tumor growth of both xenograft models following intraperitoneal administration (Fig. 2G and H). Of note, the efficacy of 6-thio-dG was comparable with that of dabrafenib that is a BRAF inhibitor approved by FDA to treat patients with BRAF-mutated melanoma as shown in A375 xenografts (Fig. 2H). Neither 6-thio-dG nor dabrafenib significantly decreased mouse weight in both xenograft models (Supplementary Fig. S1C and S1D). Both 6-thio-dG and dabrafenib substantially inhibited the protein expression of Ki-67 and phospho-S6 in A375 xenografts (Fig. 2I). The H&E staining of liver tissues from mice bearing A375 xenografts treated with 6-thio-dG or dabrafenib did not show any signs of toxicities (Fig. 2I). Taken together, our data highlight the superior antitumor activity and specificity of 6-thio-dG as seen both in vitro and in vivo.

The short-term treatment of A375 melanoma cells with 6-thio-dG but not BIBR 1532 (a telomerase inhibitor) significantly induced apoptosis and cell death (Fig. 3A). To gain mechanistic insights into the antitumor activity of 6-thio-dG, we treated A375 cells with the control, BIBR 1532, and 6-thio-dG for 48 hours, respectively. We profiled the transcriptome with RNA sequencing (RNA-seq) and the functional proteome with reverse-phase protein array (RPPA).

The differential gene expression analysis identified genes that were significantly up- and downregulated in A375 cells treated with 6-thio-dG and BIBR 1532, respectively (Supplementary Table S2 and S3). Among them were CD274 (PD-L1) and c-Myc that showed the highest degrees of inhibition, which were not inhibited in A375 cells treated with BIBR 1532 (Fig. 3B and C).

To further unravel the signaling pathways altered in A375 cells treated with 6-thio-dG or BIBR 1532, we carried out the hypergeometric differential analysis by focusing on those significantly downregulated genes. This analysis demonstrated that BioCarta gene sets related to “cell cycle” and “telomere biology” were among the top 10 ranked (Fig. 3D and E). Next, we utilized a complementary approach, single-sample gene set enrichment analysis (ssGSEA) and demonstrated that a broad collection of all of 11 telomere-related gene sets were indeed suppressed to a larger degree in A375 cells treated with 6-thio-dG compared with those treated with BIBR 1532 (Fig. 3F). Furthermore, we calculated ssGSEA scores of these 11 telomere-related gene sets in 18 CCLE melanoma cell lines and investigated correlations between ssGSEA scores and sensitivities of these cell lines to 6-thio-dG (Supplementary Table S4). We identified that three telomere-related gene sets were positively and significantly correlated with drug sensitivities, including “Telomere Maintenance,” “TERT Targets,” and “Telomerase Activity” (Fig. 3G–I). Not surprisingly, when TERT was depleted in A375 cells leading to a decrease in telomerase activity (Fig. 3J and K), the drug sensitivity to 6-thio-dG was significantly decreased (Fig. 3L). Taken together, our data suggested that the action mode of 6-thio-dG is dependent on telomerase activity and telomere maintenance.

Subsequently, we analyzed RPPA data and identified 30 proteins that were significantly downregulated in A375 cells treated with 6-thio-dG. This analysis revealed that three major signaling pathways were suppressed in A375 cells treated with 6-thio-dG. They were (i) cell cycle: phospho-RB^Ser807/811, cyclin B1, CDK1, FOXM1, PLK1 and AURKB; (ii) DNA damage response: 53BP1, ATM, ATR, CHK1 and CHK2; and (iii) receptor tyrosine kinase signaling: VEGFR, PDGFRb, IGF-IRb, IGFBP2, phospho-STAT3^Y705 and phospho-HER3^Y1289 (Fig. 3M). We also validated most of the downregulated proteins by Western blotting and showed that these proteins were indeed decreased in A375 cells treated with 6-thio-dG but not BIBR 1532 or another telomerase inhibitor, GRN163L (imetelstat; Fig. 3N and O; ref. 21). Conversely, the analysis also identified 30 proteins that were significantly upregulated in A375 cells treated with 6-thio-dG. The major signaling pathways that were upregulated were: (i) cell cycle: p53 and p21; (ii) apoptosis: Bax, caspase-3, and cleaved caspase-7; (iii) AKT signaling: phospho-AKT^T308 and phospho-AKT^S473 (Supplementary Table S5).

Because both analyses of RNA-seq and RPPA data suggested that the cell-cycle pathway is altered, we exploited A375^Fucci2 cells expressing the “fluorescent, ubiquitination-based, cell-cycle indicators Fucci, version 2” reporter construct to visualize cell-cycle progression when cells were treated with 6-thio-dG. Time-lapse imaging showed that A375^Fucci2 cells treated with 6-thio-dG over 42 hours were arrested at the G₂–M checkpoint followed by cell death (Supplementary Movie S1 and S2).

BRAF inhibitors (BRAFi) or the combination of BRAF inhibitors plus MEK inhibitors (MEKi) exhibit superior antitumor activity in BRAF-mutated melanoma cells but the development of acquired resistance inevitably occurs. This presents a great challenge in overcoming therapy resistance. To address this, we established a large panel of human melanoma cell lines that acquired resistance to BRAFi, MEKi, or the combination of BRAFi plus MEKi (19, 20). First, we asked whether therapy-resistant melanoma cells were sensitive to 6-thio-dG. Treatment of ten melanoma cell lines that had acquired resistance to PLX4720 and one melanoma cell line, WM1366 MEKi-resistant (MR) that acquired resistance to a MEK inhibitor (MEK162) with 6-thio-dG for 12 days showed a general sensitivity to 6-thio-dG and reduced viability (Fig. 4A; Supplementary Fig. S2A; Supplementary Table S6; refs. 19, 20). Next, we demonstrated that the short-term treatment of LOX-IMVR BR cells with 6-thio-dG significantly and rapidly induced apoptosis as well as cell death in the cells (Fig. 4B). Treatment of LOX-IMVI BR cells with 6-thio-dG for 120 hours led to progressive telomere shortening in residual cells (Fig. 4C). Compared with the control, 6-thio-dG also induced DNA damage (Fig. 4D–F) and led to an increase in telomere dysfunction-induced foci (TIF) that can be observed as colocalization of a specific γH2AX antibody with an in situ telomere-specific hybridization probe (Fig. 4D, E, and G).

Having demonstrated the efficacy of 6-thio-dG in treating targeted therapy-resistant melanoma cell lines in vitro, we further showed that 6-thio-dG adminstered intraperitoneally daily significantly impaired the in vivo growth of four xenografts derived from LOX-IMVI BR (Fig. 4H), WM9 BR, UACC-903 BR, and A2058 CR cells that are resistant to the BRAFi or the combination of BRAFi plus MEKi (Supplementary Fig. S2B–S2D).

We then carried out RNA-seq to profile the transcriptome of LOX-IMVI BR cells treated with 6-thio-dG. Using the ssGSEA approach, we demonstrated that all of 11 telomere-related gene sets were suppressed to a larger degree in LOX-IMVI BR cells treated with 6-thio-dG compared with those treated with BIBR 1532, which is in line with data based on A375 cells treated with 6-thio-dG (Supplementary Fig. S2E).

Next, we conducted RPPA experiments to analyze LOX-IMVI parental and BR melanoma cells that were grown in vitro or in vivo and treated with 6-thio-dG (Fig. 4I–L; Supplementary Tables S7–S10). The analysis of RPPA data of LOX-IMVI therapy-naïve and resistant cells revealed a number of proteins that were significantly upregulated in LOX-IMVI BR cells, including phospho-FAK^Y397, BRD4, EGFR, phospho-STAT3^Y705, phospho-YAP^S127, A-RAF, AXL, PI3K p85, and β-Catenin, many of which were reported to cause resistance to targeted therapy (Fig. 4I; Supplementary Table S7). Similar to A375 cells treated with 6-thio-dG, the analysis of RPPA data of LOX-IMVI BR cells showed that, three major signaling pathways were altered in LOX-IMVI BR cells treated with 6-thio-dG. They were (i) cell cycle: phospho-RB^Ser807/811, cyclin B1, CDK1, FOXM1, PLK1, AURKB, Cdc25C, and p-Cdc2^Y15; (ii) DNA damage response: 53BP1, ATM, and CHK1; (iii) receptor tyrosine kinase signaling: VEGFR and PDGFRβ (Fig. 4K; Supplementary Table S9). In addition, other functionally important proteins were also downregulated, including AXL, phospho-YAP^S127, phospho-FAK^Y397, phospho-4E-BP1^S65, and β-catenin (Fig. 4K; Supplementary Table S9). It is of particular interest to note that two proteins, AXL and phospho-YAP^Ser127, were significantly upregulated in LOX-IMVI BR cells compared with the control cells and downregulated in LOX-IMVI BR cells or xenografts treated with 6-thio-dG (Fig. 4K and L; Supplementary Table S9 and S10; ref. 25). RPPA profiling of another panel of four MAPKi-resistant cell lines treated with 6-thio-dG revealed changes at the protein level that were similar to those observed in LOX-IMVI BR cells, including the significant downregulation of AXL (Supplementary Fig. S2F–S2I). The Western blotting and IHC analyses further confirmed the downregulation of AXL in LOX-IMVI BR cells (Supplementary Fig. S2J) and A375 xenografts (Supplementary Fig. S2K) treated with 6-thio-dG, respectively.

Starting with A375^Fucci2 cells, we also established A375^Fucci2-BR and A375^Fucci2-CR cells that acquired resistance to the BRAFi and the combination of the BRAFi plus the MEKi, respectively. Not surprisingly, we found that treatment of A375^Fucci2-BR or A375^Fucci2-CR cells with 6-thio-dG also led to the cell-cycle arrest and the induction of cell death as demonstrated by the time-lapse imaging (Supplementary Movie S3–S6).

Next, we extended this study from the investigation of established human melanoma cell lines to the analysis of RNA-seq data of paired pre-, on-, and posttreatment tumor biopsies derived from patients with metastatic melanoma who progressed on targeted therapies or immunotherapies. First, we analyzed RNA-seq data derived from patients with advanced BRAF-mutated melanoma whose tumors progressed on molecularly targeted therapies. The ssGSEA approach identified “packaging of telomere ends” and “telomere maintenance” as being highly enriched in a subset of posttreatment tumor biopsies procured at the time of disease progression on BRAFi or the combination therapy of BRAFi plus MEKi compared with paired pretreatment tumor biopsies (Fig. 5A; Supplementary Fig. S3A–S3C).

Next, to investigate whether the activation of telomere signaling axis also mediates resistance to immune checkpoint blockade therapies, we first analyzed RNA-seq data of 14 posttreatment tumor biopsy specimens derived from patients with metastatic melanoma treated with ipilimumab at Memorial Sloan Kettering Cancer Center (New York, NY; refs. 26, 27). Specifically, we compared tumor biopsy specimens derived from patients experiencing “no clinical benefit” to those derived from patients experiencing “long-term clinical benefit.” We identified genes that were differentially expressed between “no clinical benefit” and “long-term clinical benefit” subgroups and then carried out GSEA to identify pathways that were associated with the phenotype of “no clinical benefit.” Interestingly, a highly ranked gene set is the “BioCarta TEL pathway” that is comprised of genes related to telomeres, telomerase, cellular aging, and immortality (Fig. 5B).

Finally, we carried out RNA-seq of paired pre-, on-, and/or posttreatment tumor biopsies derived from 16 patients with metastatic melanoma who were treated with various immune checkpoint blockade therapies at Massachusetts General Hospital (Boston, MA). The best response for most of those patients was progressive disease. The ssGSEA identified “packaging of telomere ends” and “telomere maintenance” being highly enriched in 8 out 16 patients' on- or posttreatment tumor biopsies (Supplementary Table S7). For example, “packaging of telomere ends” was highly enriched in 5 out of 8 early on-αCTLA-4 tumor biopsies (Fig. 5C), 5 out of 10 early on-αPD1 tumor biopsies (Fig. 5D), 1 out of 3 early on- αPD-L1 tumor biopsies (Fig. 5E), 4 out of 7 post-αCTLA-4 tumor biopsies (Fig. 5F), and 2 out of 2 post-αPD1 tumor biopsies (Fig. 5G).

In summary, this computational analysis reveals the reactivation of telomere signaling axis in on- and/or posttreatment tumor specimens derived from patients treated with immune checkpoint blockade therapy. This suggests that a viable approach of targeting the aberrant telomere signaling in combination with αCTLA-4, αPD-1, and/or αPD-L1 therapies could be considered.

Having demonstrated the efficacy of 6-thio-dG in inhibiting in vivo growth of MAPKi-resistant tumors and established the reactivation of telomere signaling axis in on- or posttreatment tumor biopsies derived from patients treated with immune checkpoint blockade therapies, we then examined the ability of 6-thio-dG to inhibit both in vitro and in vivo growth of melanoma cells derived from tumor specimens that are resistant to immune checkpoint inhibitors. First, we established two short-term primary cultures, T708-13-456-3-3 and T708-13-456-5-3, from two distinct BRAF^V600E-positive melanoma breast metastases that were surgically removed from a patient who progressed on sequential therapies including radiotherapy, ipilimumab, temozolomide, and pembrolizumab. Second, we established a short-term primary culture, 15-1761-1-2, from a left axillary lymph node metastatic melanoma harboring the NRAS^Q61R-positive mutation from a patient who first progressed on pembrolizumab and subsequently on the combination of ipilimumab plus nivolumab. Third, we established PDXs from two NRAS^Q61R-positive brain metastases derived from a patient with metastatic melanoma and subsequently established two PDXs-derived cell lines, WM4265-1 and WM4265-2. This patient progressed on sequential therapies, including cisplatin, vinblastine, temozolomide, IL2, IFNα-2b, ipilimumab, and pembrolizumab. In addition to therapy-resistant human melanoma, we had access to two derivatives of the mouse melanoma cell line B16: (i) 499 cell line was derived from B16 mouse melanoma that was resistant to radiotherapy and anti-CTLA-4; and (ii) JB2 cell line that differs from 499 in which PD-L1 was knocked out (28). Finally, we extended our work to nonmelanoma mouse cancer cells that acquired resistance to immune checkpoint blockade therapy to evaluate the efficacy of 6-thio-dG more broadly. This mouse pancreatic cancer cell line G43 was established from mouse pancreatic tumors harboring mutations in KRAS and P53 that progressed on radiotherapy and subsequently ipilimumab.

We treated 13-456-3-3, 13-456-5-3, WM4265-1, WM4265-2, 499, JB2, and G43 cells with 6-thio-dG and observed that prolonged treatment with 6-thio-dG for 12 days markedly inhibited in vitro cell proliferation (Fig. 6A). Furthermore, we showed impressive antitumor activity of 6-thio-dG in six xenografts, including 13-456-3-3 (Fig. 6B), 13-456-5-3 (Fig. 6C), 15-1761-1-2 (Fig. 6D), WM4265-2 (Fig. 6E), 499 (Fig. 6F), and G43 (Fig. 6G) when administered intraperitoneally daily.

Using 13-456-5-3 cells and xenografts as a representative pair that was treated with 6-thio-dG in vitro and in vivo, respectively, we conducted the RPPA experiment and identified proteins that were significantly downregulated by 6-thio-dG. The analysis identified 9 proteins that were commonly downregulated, including phospho-Myosin^S1943, ARID1A, BRD4, AXL, ATM, eIF4G, PLK1, SOX2, and 53BP1 (Fig. 6H and I). We further extended the RPPA analysis to 5 therapy-resistant cell lines that were treated with 6-thio-dG in vitro, including 13-456-3-3, 15-1761-1-2, WM4265-1, WM4265-2, and G43. Importantly, AXL was downregulated in all cell lines except G43 that were treated with 6-thio-dG (Supplementary Fig. S5A–S5E).

In summary, we demonstrated that 6-thio-dG significantly impaired the growth, both in vitro and in vivo, of melanoma cells derived from various models that acquired resistance to immune checkpoint blockade therapies, implying that 6-thio-dG could be used as a viable therapeutic approach for therapy-resistant melanomas.

Following the initial phase of therapeutic responses, patients with unresectable or metastatic melanomas frequently relapse on MAPKi targeting the BRAF^V600mutations. Molecular mechanisms of acquired resistance to MAPKi are heterogeneous and complex, necessitating the genetic dissection of the mutational spectrum of therapy-resistant tumors to personalize second-line therapies for each individual patient.

Although immune checkpoint blockade therapies that target CTLA4, PD-1, or PD-L1 can elicit durable responses, the majority of patients do not experience durable clinical benefit. For instance, a phase Ib clinical trial demonstrated that the objective response rate was 33% and 45%, respectively, for all patients and treatment-naïve patients treated with pembrolizumab (29). Current studies demonstrate that the lack of response to immune checkpoint blockade therapies is, at least in part, due to expression of the PD-1/PD-L1 immune inhibitory axis, activation of alternative immune checkpoints such as TIM-3, and activation of a melanoma-intrinsic WNT/β-catenin signaling axis, resulting in T-cell exclusion and resistance (30–32).

Because TERT promoter mutations are highly prevalent in melanomas resulting in high telomerase activity and molecular mechanisms underlying therapy resistance are heterogeneous, we hypothesized that targeting telomerase activity in therapy-resistant and telomerase-positive tumors might provide a universal strategy to achieve long-term and stable melanoma control.

To therapeutically target telomerase-positive melanomas, we exploited a novel telomere-directed, telomerase-dependent anticancer inhibitor, 6-thio-dG, in a panel of BRAF-mutated human melanoma cell lines. Telomerase recognizes the nucleoside 6-thio-dG, which is subsequently incorporated into de novo–synthesized telomeres, leading to DNA damage localized at telomeric DNA regions and the rapid killing of cancer cells. Previously, we demonstrated that 6-thio-dG did not lead to weight loss or changes in hematologic, renal, or liver function at effective in vivo doses, which was further confirmed by this study (18). Similar to other signaling inhibitors such as the BRAFi, responses of BRAF-mutant melanoma cell lines to 6-thio-dG are heterogeneous, although the vast majority was sensitive.

We compared the in vitro efficacy of 6-thio-dG to that of PLX4720 and demonstrated 6-thio-dG alone is able to elicit apoptosis and cell death in BRAF-mutated melanoma cell lines, including ones that are intrinsically resistant to PLX4720. Extended treatment with 6-thio-dG impairs cell viability and induces cellular senescence in the small fraction of surviving cell, resulting in progressively shortened telomeres, mimicking a phenotype of replicative senescence exhibited by normal cells. We also compared the in vivo efficacy of 6-thio-dG to that of another BRAF inhibitor, dabrafenib, and demonstrated that both of them had very similar effect in inhibiting the tumor growth.

To investigate alterations in signaling transduction proteins that are induced by 6-thio-dG, we conducted integrated computational analyses of melanoma cells treated with 6-thio-dG using RNA-seq and RPPA platforms. We discovered that cell cycle, telomere signaling, DNA damage response and molecules causing therapy resistance are decreased in cells treated with 6-thio-dG.

We speculate that short-term treatment with MAPKi might select a subpopulation of BRAF-mutated melanoma cells in which AXL and other resistant markers are upregulated and telomere transcriptional gene signatures are enriched, allowing them to adapt to, and evade MAPKi. The “one-two punch” ability of 6-thio-dG to downregulate resistant markers such as AXL among others at the protein level, and induce telomere dysfunction makes it an attractive approach with the potential to overcome both intrinsic and acquired drug resistance to MAPKi. In this study, we demonstrated that 6-thio-dG is able to inhibit the growth of melanoma cells that acquired drug resistance to MAPKi both in vitro and in vivo. This warrants a future study in which we explore the efficacy of combining MAPKi with 6-thio-dG in overcoming both intrinsic and acquired resistance to MAPKi.

Only a subset of patients with advanced and metastatic melanomas experience long-term and durable clinical benefits from immune checkpoint blockade therapies. To enhance benefit from immune checkpoint blockade, it is critical to dissect and therapeutically target molecular mechanisms that underlie adaptive and acquired resistance to immune checkpoint blockade therapies. By analyzing unpaired pre- and posttreatment tumor biopsy specimens derived from patients who did not benefit from ipilimumab as well as paired pre-, on-, and posttreatment tumor biopsy specimens derived from patients who did not benefit from immune checkpoint blockade therapies, we identified “telomeres, telomerase, cellular aging, immortality,” “telomere maintenance” and “packaging of telomere ends,” that are enriched in on- and/or posttreatment tumor biopsies. Collectively, our data support the notion that the reactivation of the telomere signaling axis could confer a survival advantage upon tumors that are refractory to immune checkpoint blockade therapies.

Having shown that 6-thio-dG is able to prolong the disease control in multiple preclinical models of melanoma cells that acquired resistance to MAPKi, we then hypothesized that 6-thio-dG might also be able to prolong disease control in preclinical models of melanoma cells that acquired resistance to immune checkpoint blockade therapies. Importantly, we demonstrated activity of 6-thio-dG in impairing the growth of short-term primary cultures in vitro and in vivo that are established from metastatic lesions derived from patients who progress on multiple therapies including ipilimumab and pembrolizumab.

In summary, our current results are highly supportive of using 6-thio-dG as monotherapy following first-line or second-line therapies to prolong disease control of human therapy–resistant and telomerase-positive tumors. This study also paves the way for exploring the combination of 6-thio-dG and first-line therapies in overcoming both intrinsic and acquired resistance.

L.N. Kwong reports receiving a commercial research grant from Array Biopharma. G.B. Mills reports receiving a commercial research grant from Adelson Medical Research Foundation, AstraZeneca, Critical Outcome Technologies, Komen Research Foundation, NanoString, Breast Cancer Research Foundation, Karus, Illumina, Takeda/Millenium Pharamceuticals, Pfizer, has received speakers bureau honoraria from Symphogen, MedImmune, AstraZeneca, ISIS Pharmaceuticals, Lily, Novartis, ImmunoMet, Allostery, Tarveda, Pfizer, has ownership interest (including patents) Catena Pharmaceuticals, PTV Ventures, Spindletop Ventures, Myriad Genetics, ImmunoMet, and is a consultant/advisory board member for Adventist Health, AstraZeneca, Provista Diagnostics, Signalchem Lifesciences, Symphogen, Lilly, Novartis, Tarveda, Tau Therapeutics, Allostery, Catena Pharmaceuticals, Critical Outcome Technologies, ISIS Pharmaceuticals, ImmunoMet, Takeda/Millenium Pharmaceuticals, MedImmune, and Precision Medicine. J.W. Shay has ownership interest (including patents) in Barricade Therapeutics. No potential conflicts of interest were disclosed by the other authors.

Conception and design: G. Zhang, X. Xu, G.B. Mills, J.W. Shay, M. Herlyn

Development of methodology: G. Zhang, O. Ope, J. Tan, X. Xu, J.W. Shay

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Zhang, L.W. Wu, I. Mender, M. Barzily-Rokni, M.R. Hammond, O. Ope, N. Sadek, M. Xiao, U. Saeed, E. Sugarman, C. Krepler, P.A. Brafford, K. Sproesser, S. Murugan, R. Somasundaram, J. Woo, D.T. Frederick, W. Xu, G.C. Karakousis, X. Xu, L.M. Schuchter, T.C. Mitchell, L.N. Kwong, G. Boland, K.L. Nathanson, G.B. Mills, K.T. Flaherty, T. Vasilopoulos

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Zhang, L.W. Wu, I. Mender, O. Ope, C. Cheng, S. Randell, T. Tian, U. Saeed, E. Sugarman, C. Krepler, R. Somasundaram, B. Wubbenhorst, J. Woo, X. Yin, Q. Liu, R.K. Amaravadi, G. Boland, Z. Wei, K.L. Nathanson, G.B. Mills, K.T. Flaherty, J.W. Shay, M. Herlyn

Writing, review, and/or revision of the manuscript: G. Zhang, L.W. Wu, O. Ope, S. Randell, N. Sadek, U. Saeed, E. Sugarman, Q. Liu, W. Xu, G.C. Karakousis, X. Xu, T.C. Mitchell, R.K. Amaravadi, Z. Wei, K.L. Nathanson, K.T. Flaherty, J.W. Shay, M. Herlyn

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G. Zhang, L.W. Wu, A. Beroard, U. Saeed, J. Woo, B. Miao, W. Xu, R.K. Amaravadi, K.L. Nathanson, T. Vasilopoulos, U. Herbig

Study supervision: G. Zhang, Z. Wei, J.W. Shay

Other (conducted experiments): B. Garman

Other (performed RPPA assay and provided raw data): Y. Lu

The research was funded by NIH grants P01CA114046, P50CA174523, P50CA70907, 1U54CA224070, DoD PRCRP grant CA150619, AG01228, the Southland Financial Corporation Distinguished Chair in Geriatric Research, the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, and the Melanoma Research Foundation; NIH grant 5R01CA136533 to U. Herbig; and by the Tara Miller Foundation to L.M. Schuchter. The support for Shared Resources utilized in this study was provided by Cancer Center Support Grant (CCSG) CA010815 to The Wistar Institute, CA016672 to the MDACC, and CA142543 to UTSW. Part of this work was performed in laboratories constructed with support from NIH grant C06 RR30414. The authors thank all former and current lab members for comments and helpful discussions; J. Hayden and F. Keeney (Wistar Microscopy Facility), C. Chang, S. Billouin, and T. Nguyen (Wistar Genomics Facility), J.S. Faust (Wistar Flow Cytometry Facility), D. DiFrancesco (Wistar Animal Facility), and F. Chen (Wistar Histotechnology Facility) for technical support; and M.B. Powell for providing human melanoma cells (Stanford University). The authors apologize to those whose work was not cited or mentioned here due to space constraints.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.