Telomere damage enhances immunogenicity of neuroblastoma and accelerates response to anti-PD-L1 treatment

Abstract

Telomerase is active in the majority of high-risk neuroblastomas, a pediatric tumor associated with poor patient outcomes. In other cancer types, resistance to immune checkpoint blockade was overcome by induction of telomere dysfunction using the telomerase substrate precursor 6-thio-2ʹ-deoxyguanosine (6-thio-dG). Here, we explored whether induction of telomere dysfunction improves the anti-tumor efficacy of immune checkpoint inhibition in neuroblastoma. 6-thio-dG treatment induced the cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathway and programmed cell death ligand-1 (PD-L1) expression in murine neuroblastoma cells in vitro. In a MYCN;ALK^F1174L-driven transgenic neuroblastoma mouse model, 6-thio-dG treatment delayed tumor growth and prolonged survival. Treatment with anti-PD-L1 also led to growth delay and improved survival, which occurred; however, only after an initial lag phase. Combination with anti-PD-L1 improved the anti-tumor effects of 6-thio-dG and overcame the initial lag phase of anti-PD-L1 treatment. Mechanistically, 6-thio-dG combined with anti-PD-L1 treatment induced cGAS and PD-L1 expression and promoted immune cell infiltration in the tumors. Our findings suggest that 6-thio-dG treatment activates the cGAS-STING pathway in neuroblastoma and that induction of telomere dysfunction in combination with immune checkpoint blockade boosts intratumoral immune cell infiltration and improves survival in a high-risk neuroblastoma mouse model.

Introduction

Neuroblastoma originates from the developing sympathetic nervous system and is the most common extracranial solid tumor in children. It accounts for 8%–10% of pediatric cancers and approximately 15% of cancer-related pediatric deaths.¹ According to the International Neuroblastoma Risk Group (INRG) classification, high-risk neuroblastoma is defined by metastatic disease in children older than 18 months (stage M) or presence of MYCN amplification in the tumor cells.² High-risk patients still have poor outcome, with a 10 y overall survival of approximately 50%, despite intensive treatment involving chemotherapy, surgery, autologous stem cell transplantation, irradiation, and anti-GD2 immunotherapy.^3-7 Thus, novel therapeutic strategies targeting high-risk neuroblastoma are urgently needed.

Activation of telomere maintenance mechanisms (TMM) has been reported as a hallmark of high-risk neuroblastoma, whereas such alterations were absent in low-risk disease.^8,9 TMM is conferred either by induction of telomerase or alternative lengthening of telomeres (ALT).^8,9 Since telomerase is expressed almost exclusively in malignant cells, it may present a promising target for telomerase-positive high-risk neuroblastoma treatment. Telomerase, a ribonucleoprotein enzyme consisting of the catalytic telomerase reverse transcriptase (TERT) and an RNA template (TERC), elongates the telomeres of the chromosomes, which enables cancer cells with immortal proliferation capacity.¹⁰ Constitutive TERT activation occurs in roughly two-thirds of high-risk neuroblastoma, and may be caused by transcriptional induction through enhancer hijacking events at the TERT locus or amplified MYCN.¹¹

It has been demonstrated recently that the nucleoside analog 6-thio-2ʹ-deoxyguanosine (6-thio-dG) is incorporated into telomeres by telomerase, resulting in telomere uncapping, telomere-associated DNA damage, and subsequent tumor cell death, thus providing a therapeutic strategy to target telomerase-positive cancers.¹² In preclinical studies on neuroblastoma, 6-thio-dG extended the survival of mice bearing human neuroblastoma xenografts with high TERT expression.¹³ In addition to direct cytotoxic effects, telomere dysfunction leads to the release of DNA fragments that activate the cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathway.¹⁴ Cytosolic DNA fragments bind cGAS, which catalyzes cGAMP production and activates STING.¹⁵ This, in turn, stimulates TANK-binding kinase 1 (TBK1) and phosphorylates interferon regulatory factor 3 (IRF-3), thereby driving type I interferon production and initiating an innate immune response.¹⁶ The activated innate immune system is critical for engaging adaptive immunity, which is essential for robust anti-tumor responses.¹⁷ In preclinical in vivo studies on lung adenocarcinoma and colon carcinoma, treatment with a combination of 6-thio-dG and anti-PD-L1 antibodies overcame T cell exhaustion by upregulation of the checkpoint molecules PD-1/PD-L1, thus leading to a strong anti-tumor response.¹⁶ This approach is currently being tested in an ongoing phase 2 clinical trial in advanced non-small cell lung carcinoma (NCT05208944).¹⁸

Neuroblastoma is considered an immunologically “cold” tumor.¹⁹ While adult cancers most frequently originate from epithelia and develop as a consequence of continuous DNA damage, resulting in a high mutational burden and a pro-inflammatory tumor microenvironment,²⁰ pediatric tumors often arise from few genetic changes in cells of embryonic or mesenchymal origin with reduced MHC-I expression.²¹ Thus, little immune cell infiltration and limited immune recognition of tumor cells have been found in pediatric cancers.^22,23 In neuroblastoma, immune cell infiltration, including B cells, T cells, and NK cells, occurs primarily in low-risk tumors, whereas it is limited in high-risk tumors.¹⁹ In addition, cytotoxic CD8+ T cells, crucial for targeting cancer cells, are often dysfunctional or exhausted in neuroblastoma, partly due to expression of the immune checkpoint protein PD-L1.¹⁹ In line with these observations, high PD-L1 expression is associated with high-risk disease and poor outcome in neuroblastoma.^24,25 The potential clinical impact of checkpoint inhibition using anti-PD-L1 antibodies²⁶ in neuroblastoma, however, has remained unclear. While preclinical in vivo studies targeting the PD-L1/PD-1 axis in neuroblastoma mouse models,^27,28 as well as first clinical approaches with anti-PD-1/anti-PD-L1 monotherapy in neuroblastoma patients,^29,30 did not result in survival benefits, response to combined anti-PD-1 and anti-GD-2 therapy has been reported in two refractory neuroblastoma patients.³¹

We set out to test the hypothesis that induction of telomere dysfunction sensitizes neuroblastoma to immune checkpoint inhibition through activation of the cGAS-STING pathway. To this end, we determined the anti-tumor effects of treatment with 6-thio-dG and anti-PD-L1 alone and in combination, as well as their molecular consequences on the cGAS-STING pathway in an immunocompetent transgenic neuroblastoma mouse model. We demonstrate that 6-thio-dG can prime neuroblastoma for anti-PD-L1 treatment by inducing immune cell infiltration and reducing the lag period preceding PD-L1 treatment response in a model of aggressive high-risk neuroblastoma.

Methods

Cell culture

The NHO2A cell line, derived from the transgenic Th-MYCN neuroblastoma mouse model³² and kindly provided by Alexander Schramm, was cultured in a humidified incubator at 37 °C with 5% CO₂. THP-1 cells were used as a human positive control for expressing high levels of cGAS. All cells were maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. For treatment experiments involving the telomerase inhibitor 6-thio-dG, NHO2A cells were seeded at a density of 5 × 10⁵ cells/mL in 10 cm dishes. After 24 h, they were treated with 200 nM or 1 μM 6-thio-dG (Selleckchem, S7757) or DMSO as a control for 96 h. To assess the effects of STING, 2 μM of the STING antagonist SN-011 (Selleckchem, E1066) was added to the culture medium for 96 h. Additionally, NHO2A cells were treated with 20 μM of the STING agonist DMXXA (Selleckchem, S1537) and THP-1 cells with 1 µM STING agonist diABZI (Selleckchem, S8796), serving as a positive control. Cell treatment was performed in biological triplicates. The cell lines were regularly tested for mycoplasma contamination (q-PCR-based, Eurofins).

Mice

Th-MYCN and Th-ALK^F1174L mice, as previously described,^33,34 were interbred and genotyped at 14 d of age. Genotyping was performed using specific PCR primers: Chr18F1 (ACT AAT TCT CCT CTC TCT GCC AGT ATT TGC), Chr18R2 (TGC CTT ATC CAA AAT ATA AAT GCC CAG CAG) and OUT1 (GCA CAC ACA AAT GTA TAT ACA CAA TGG) for the Th-MYCN transgene, and ALK1091for (GGC ATC ATG ATT GTG TAC CG) and ALK1174rev (ATG AGC TCC AGC AGG ATG AA) for the Th-ALK ^F1174L allele. The animals were housed in a specific-pathogen-free facility under controlled conditions, with groups of up to five mice per cage, standard pellet food, and water provided ad libitum. The housing environment was maintained on a 12-h light/dark cycle with temperatures controlled between 21 and 22 °C. All experiments were conducted in accordance with FELASA recommendations, the ARRIVE guidelines, and were approved by the local animal care committee and relevant authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, AZ: 81-02.04.2021.A395).

Mouse experiments

A total of 122 males and females were chosen by random draw for different treatment cohorts and treated between January 2022 and May 2024. Mice were treated with a combination of 6-thio-dG and anti-PD-L1 antibody, with 6-thio-dG or anti-PD-L1 as monotherapies, and compared to control groups receiving 6-thio-dG plus IgG control, IgG control alone, or NaCl 0.9%. Additionally, the combination of 6-thio-dG and anti-PD-L1 was directly compared to 6-thio-dG and anti-PD-L1 monotherapy. The treatment regimen for mice included intraperitoneal injections of 6-thio-dG (Selleckchem, S7757) 2.5 mg/kg on days 1, 2, and 3 during the first treatment cycle (24 d), followed by the same schedule on days 25, 26, and 27 in the second cycle (24 d). Concurrently, PD-L1 (BioXcell, B7-H1, clone 10F.9G2, 200 μg/mouse) or isotype control IgG (BioXcell, BP0090, LTF-2, 200 μg/mouse) treatments were administered on days 6, 9, 12, 18, 21, and 24 in the first cycle and on days 30, 33, 36, 42, 45, and 48 in the second cycle. Mice in control groups received NaCl 0.9% (10 ml/kg) on corresponding days for each cycle. After completing the second cycle, treatments were restarted following the schedule of cycle one and continued until mice reached humane endpoints in accordance with the scoring protocol. Confounding factors were not controlled, and blinding of investigators to treatment groups was not possible. A total of 22 mice were treated in the 6-thio-dG + anti-PD-L1 group, while 20 mice were included in each of the remaining groups. For survival analyses, each individual mouse was considered an experimental unit, whereas treatment response was evaluated based on individual tumor size measurements (see below). Mice were sacrificed by cervical dislocation upon reaching predefined human endpoints (poor general appearance, weight loss, apathy, e.g.) in accordance with the American Veterinary Medical Association, and tumors were collected accordingly. Tumors were either snap-frozen or embedded in paraffin for further analyses. The following parameters were assessed: Tumor growth via weekly MRI scan, survival rates, tumor immune cell infiltration, and activation of the cGAS-STING pathway.

Magnetic resonance imaging

Initial magnetic resonance imaging (MRI) scans were conducted at three weeks of age, with weekly follow-up scans. MRI was performed on a clinical 3.0 T MRI system (Achieva Quasar Dual, Philips Healthcare, Netherlands) using a small rodent solenoid coil with a 40 mm inner diameter and a heating system to maintain body temperature during imaging (Philips Research Europe, Hamburg, Germany). Anesthesia was induced with 2.0%–2.5% isoflurane inhalation. High-resolution axial T2-weighted images were acquired using a multishot turbo-spin echo (TSE) sequence, with repetition times of 7639 ms (axial), echo time of 60 ms, flip angle of 90°, and a reconstructed resolution of 0.16 × 0.16 mm² with 1 mm slice thickness. Images were exported in DICOM format and analyzed using Horos software (Version 3.3.6). MRI scans were analyzed in a blinded manner with respect to treatment groups. Tumor size was assessed by measuring the maximum axial diameter of each tumor individually. The relative tumor diameter was calculated by normalizing each measurement to the largest recorded diameter of the respective tumor. The average tumor growth dynamics and standard deviations of relative tumor diameters were then determined across treatment groups.

Immunohistochemistry on fresh-frozen tissue sections

Immunohistochemistry was conducted on fresh-frozen tumor sections for 10 tumors from each treatment group. Five-micrometer sections were dried on superfrost slides for 30 min and fixed in cold acetone at −20 °C for 10 min. After rehydrating in PBS for 10 min, endogenous peroxidase was blocked with 10% hydrogen peroxide in methanol at room temperature. Slides were washed in PBS, blocked with 5% normal goat serum in PBS containing 0.3% Triton X-100 for 1 h, and incubated overnight at 4 °C with primary antibodies diluted in blocking solution. The following antibodies were used: CD45 (1:500, abcam, ab10558), CD4 (1:500, abcam, ab237722), CD8 (1:500, abcam, ab217344), CD19 (1:250, abcam, ab245235), and CD11b (1:1000; abcam, ab1338357). Signals were detected using SignalStain® Boost IHC Detection Reagent (HRP, rabbit, Cell Signaling) for 30 min, followed by SignalStain® DAB substrate (cell signaling) for 10 min at room temperature. After counterstaining with hematoxylin, slides were mounted with AquaTex (Sigma Aldrich) and scanned at 40× magnification using a NanoZoomer S360 slide scanner (Hamamatsu Photonics). For quantification of immune cells, tumors were analyzed using OMERO software (Version 5.18.0). From each treatment group, one representative tumor was chosen for further analysis. From each tumor, five randomly chosen fields of view (each comprising 0.62 mm²) were selected, and CD45, CD4, CD8, CD19, and CD11b positive cells were counted using the FIJI (version 2.16.0/1.54p) plugin cell counter (version 3.0.0).

Histopathology

For histopathology, tumors and organs were harvested and fixed in 4% PBS-buffered formalin before paraffin embedding. Three-micrometer sections were deparaffinized and stained with hematoxylin and eosin following standard diagnostic protocols provided by the Institute of Pathology at University Hospital Cologne.

Patient samples

Patient samples for western blot analyses (n = 5) and sequencing (n = 10) were obtained from patients included in the clinical trials NB97 (NCT00017225, https://clinicaltrials.gov/study/NCT00017225#collaborators-and-investigators), NB2004 (NCT03042429, https://clinicaltrials.gov/study/NCT03042429) and NB2016 Registry (DRKS00023442, https://drks.de/search/de/trial/DRKS00023442) of the Gesellschaft für Pädiatrische Onkologie und Hämatologie (GPOH) with informed consent by patients or their legal guardians. The trials were performed in accordance with the principles stated in the Declaration of Helsinki. To compare CGAS expression between neuroblastoma cell lines and neuroblastoma tumors, we used previously published sequencing data (SEQC cohort).³⁵ Additionally, we sequenced tumor material from patients under treatment (n = 10) and compared expression levels of CGAS, TMEM173 (STING) and CD274 (PD-L1) with non-matched initial high-risk samples of the SEQC cohort (n = 176). Five tumor samples from patients with high-risk disease were chosen for western blot analyses.

The use of tumor material for research was approved by the institutional ethical review board of the University of Cologne (vote 23-1387).

Western blot

Western blot analysis was conducted by lysing cells in RIPA buffer (Thermo Fisher Scientific) supplemented with Protease and Phosphatase Inhibitor Cocktail (PPI, Life Technologies). Tumor tissue samples (20 μm sections) were sonicated on ice in RIPA buffer plus PPI (10 mg tumor in 100 µl RIPA + PPI) using an ultrasonic processor (Hielscher Ultrasonic, UP200St). Protein concentrations were measured using a Bradford assay. Proteins were separated and transferred to PVDF membranes using standard immunoblotting procedures. Primary antibodies used included PD-L1 (Cell Signaling, #D4H1Z (mouse) or #E1L3N (human); 1:1000), cGAS (Cell Signaling, #D3080 (mouse) or #D1D3G (human); 1:1000), IRF-3 (Cell Signaling, #D83B9; 1:1000), phosho-IRF-3 (Cell Signaling, #S396; 1:1000) and HSP-90 (Cell Signaling, #C45G5; 1:1000). Membranes were incubated with primary antibodies overnight at 4 °C and with HRP-conjugated secondary antibodies for 1 h (4 °C) and detected using a chemiluminescence detection system (Biorad).

RNA sequencing

RNA sequencing libraries were prepared and processed according to the Illumina non-stranded TruSeq® protocol, as described previously.³⁵ The raw sequencing data were quantified using Kallisto (version 0.44.0) with parameters –bootstrap -samples 100 –bias and the GRCh38 reference genome. The Kallisto transcript-level read counts were transformed to gene-level counts using the R package tximport (version 1.36.1). To obtain gene expression values, the counts were normalized with DESeq2 (version 1.48.2), and a variance-stabilizing transformation was applied. P-values were computed using a two-sided Wilcoxon rank-sum test. Boxplots indicate the median, 1st and 3rd quartiles, their whiskers the minimum and maximum values within ±1.5 times the interquartile range. Sequencing data of cell lines are accessible at GEO under accession number GSE312549.

Statistical analyses

Statistical analyses were conducted using R (Version 4.2.2) and GraphPad Prism (Version 9.5.1). Sample size calculations were performed a priori using Chi² test with the software PS power and Sample Size Calculations (Version 3.0) based on preliminary data. Kaplan–Meier survival analysis was used to assess disease-specific survival, defined as the time from birth to death, with group comparisons made using the log-rank test. All mice were included in the survival analyses; however, animals that died due to unrelated causes, not attributable to tumor progression or health condition, were censored. These exclusion criteria were predefined prior to the start of the experiments.

Tumor regression analyses were performed using Fisher's exact test, with multiple testing correction applied using the Benjamini–Hochberg method. Every regressing tumor that occurred was included in the analysis. A total of 88 tumors were analyzed in the 6-thio-dG plus anti-PD-L1 group, 92 in the 6-thio-dG plus IgG control group, 30 in the 6-thio-dG monotherapy group, 91 in the anti-PD-L1 monotherapy group, 5 in the IgG control group, and 1 in the NaCl 0.9% group.

Statistical analyses of the CD45+, CD4+, CD8+, CD19+, and CD11b+ cells per field of view in the immunohistochemical staining were carried out with R (version 4.5.1). Comparisons between treatment groups were computed with a one-sided Wilcoxon rank-sum test and corrected for multiple testing using Benjamini–Hochberg correction.

Figure generation

Figures were generated with Adobe Illustrator (Version 25.4.8) and GraphPad Prism (Version 10.1.2).

Results

Treatment with the telomere-targeting drug 6-thio-dG activates the cGAS-STING pathway and upregulates PD-L1 expression in murine neuroblastoma cells in vitro

Telomere damage is known to be sensed by the cGAS-STING pathway,¹⁴ leading to a type I interferon response and subsequent immune cell infiltration. To evaluate the direct effects of 6-thio-dG on the cGAS-STING pathway and on PD-L1 expression in neuroblastoma tumor cells in vitro, we treated NHO2A cells, a neuroblastoma cell line derived from a Th-MYCN transgenic mouse model,³² with this compound. We observed upregulation of cGAS and phosphorylation of IRF-3 (pIRF-3), a key downstream effector of STING (Figure 1a). Inhibition of STING by SN-011, a potent STING antagonist,³⁶ abolished these effects, thus confirming STING pathway activation by 6-thio-dG in these cells. In addition, we observed upregulation of PD-L1 expression following 6-thio-dG treatment, which was also reversed by SN-011, establishing a direct interaction between STING activation and PD-L1 upregulation.³⁷

Figure 1.

(a) Immunoblot analysis of NHO2A cells without treatment, treated with DMSO, or treated with 6-thio-dG at 200 nM or 1 µM for 96 h in the absence or presence of the STING antagonist SN-011 at 2 µM. Treatment with the STING agonist DMXXA (20 µM for 24 h) served as positive control; HSP-90 served as loading control. Representative blot from three independent experiments. (b) Treatment schedule of Th-MYCN;Th-ALK^F1174L mice. Vertical arrows are indicating treatment days, with green arrows for 6-thio-dG, blue arrows for PD-L1, red arrows for IgG control, and black arrows for NaCl 0.9%. Treatment cycle 1 spans days 1 to 24, cycle 2 days 25 to 48. Animals were randomly assigned to one of six cohorts, and treatment was started at the age of 3 w after tumor detection by MRI scan.

6-Thio-dG treatment extends survival and delays tumor progression in Th-MYCN;Th-Alk^F1174L transgenic mice

To evaluate the anti-tumor effect of 6-thio-dG on neuroblastoma in an immune-competent background, we used Th-MYCN;Th-Alk^F1174L transgenic mice,³³ a murine model of aggressive neuroblastoma reflecting a genotype that occurs recurrently in patients.³⁸ Mice were genotyped at the age of 14 d, and the occurrence of tumors was confirmed at the age of 3 w by MRI. Treatment with 6-thio-dG (2.5 mg/kg) was initiated immediately after tumor detection and administered at days 1, 2, and 3 of a 24-d cycle (Figure 1b). We opted for this comparatively low 6-thio-dG intensity, as we observed severe pancytopenia after several weeks of treatment with doses that have been used in previous studies^13,16 (data not shown). Weekly MRI scans were conducted to monitor tumor growth and regression over the treatment period, with the largest tumor diameters measured after each scan. The relative tumor diameter was determined by normalizing each tumor diameter to the largest recorded diameter of the corresponding tumor.

After 25 d, a second cycle of 6-thio-dG treatment was started with the same dosing regimen. Treatment cycles were iterated until the stop criteria were reached. 6-thio-dG treatment significantly improved overall survival of the treated cohort compared to mice receiving NaCl 0.9% as control (Figure 2a; p = 0.018), similar to the anti-tumor effect of 6-thio-dG on mice bearing human neuroblastoma xenograft tumors.^13,39 While the overall effect on tumor growth was limited (Figure 2b), we observed transient regression of tumors in 32% of the treated animals, which occurred predominantly within 14 d after treatment, whereas regression was observed in only 2% of mice of the control group (Figure 2c; p < 0.001). Thus, low-intensity 6-thio-dG treatment delayed tumor progression, leading to prolonged survival in this neuroblastoma mouse model.

Figure 2.

Anti-tumor effects of 6-thio-dG monotherapy or PD-L1 monotherapy and combination of 6-thio-dG with anti-PD-L1 antibody in Th-MYCN;Th-ALK^F1174L mice. (a) Kaplan–Meier estimates of survival of mice treated with 6-thio-dG (purple) or NaCl 0.9% (black). Survival curves were compared by log-rank test. (b) Average growth dynamics and standard deviation (shaded areas) of relative tumor diameters in mice treated with 6-thio-dG (purple) or NaCl 0.9% (black). The number of treated mice, tumors, as well as absolute and relative incidence of tumor regression per cohort, is indicated. (c) Probability of tumor regression in mice treated with 6-thio-dG vs. NaCl 0.9%. Groups were compared by Fisher's exact test with Benjamini–Hochberg correction for multiple testing. (d) Kaplan–Meier estimates of survival of mice treated with 6-thio-dG in combination with anti-PD-L1 antibody (green) or NaCl 0.9% (black). (e) Average growth dynamics and standard deviation (shaded areas) of relative tumor diameters in mice treated with 6-thio-dG in combination with anti-PD-L1 antibody (green) or NaCl 0.9% (black). (f) Probability of tumor regression in mice treated with 6-thio-dG in combination with anti-PD-L1 antibody vs. NaCl 0.9%. (g) Kaplan–Meier estimates of survival of mice treated with anti-PD-L1 antibody (blue) or NaCl 0.9% (black). Survival curves were compared by the log-rank test. (h) Average growth dynamics and standard deviation (shaded areas) of relative tumor diameters in mice treated with anti-PD-L1 antibody (blue) or NaCl 0.9% (black). The number of treated mice, tumors, as well as the absolute and relative incidence of tumor regression per cohort, is indicated. (i) Probability of tumor regression in mice treated with anti-PD-L1 antibody vs. NaCl 0.9%. Groups were compared by Fisher's exact test with Benjamini–Hochberg correction for multiple testing. (j) Kaplan–Meier estimates of survival of mice treated with 6-thio-dG in combination with anti-PD-L1 antibody (green) vs. anti-PD-L1 antibody alone (blue). (k) Average growth dynamics and standard deviation (shaded areas) of relative tumor diameters in mice treated with 6-thio-dG in combination with anti-PD-L1 antibody (green) or anti-PD-L1 antibody alone (blue). (l) Probability of tumor regression in mice treated with 6-thio-dG in combination with anti-PD-L1 antibody vs. anti-PD-L1 antibody alone.

Anti-PD-L1 therapy augments the anti-tumor effect of 6-thio-dG in Th-MYCN;Th-Alk^F1174L transgenic mice

To evaluate whether immune checkpoint blockade improves the therapeutic efficacy of 6-thio-dG in murine neuroblastoma, we combined 6-thio-dG with anti-PD-L1 antibody treatment. Anti-PD-L1 antibody (200 µg/mouse) was given on days 6, 9, 12, 18, 21, and 24 of each cycle, following 6-thio-dG treatment on days 1-3 (Figure 1b). This combination significantly extended the overall survival of mice compared to both 6-thio-dG monotherapy (Supplementary Figure 1a; p = 0.044) and the control group (Figure 2d; p < 0.001). Improved survival was associated with delayed tumor growth and a higher frequency of transient tumor regression in comparison to the control group (Figure 2e, f). Transient regression in the combination treatment group was also more frequent when compared to mice treated with 6-thio-dG monotherapy (Supplementary figure 1b, c).

To assess the therapeutic specificity of checkpoint inhibition using an anti-PD-L1 antibody, we treated another control group of mice with 6-thio-dG in combination with an isotype control antibody (cIgG). Survival of mice and growth characteristics of tumors treated with this combination did not differ substantially from those treated with 6-thio-dG monotherapy (Supplementary Figure 1d–f). We noted, though, that survival of mice treated with a combination of 6-thio-dG and cIgG did not differ from that of mice treated with 6-thio-dG and anti-PD-L1, although the latter had a higher fraction of transient tumor regression and slightly delayed tumor growth in the first weeks of treatment (Supplementary Figure 1g–i). Together, these findings suggest that PD-L1 blockade may have a moderate synergistic effect with the induction of telomere dysfunction in murine neuroblastoma driven by MYCN and ALK^F1174L.

Anti-PD-L1 monotherapy improves survival and attenuates tumor growth in Th-MYCN;Th-Alk^F1174L transgenic mice

Since the potential therapeutic impact of immune checkpoint blockade in neuroblastoma has remained uncertain, we also assessed the anti-tumor effect of anti-PD-L1 monotherapy. To this end, we administered anti-PD-L1 antibody on days 6, 9, 12, 18, 21, and 24 post-tumor detection, and compared it to mice treated with cIgG only or NaCl 0.9% as a control (Figure 1b). The two control groups did not differ in their survival or tumor growth characteristics (Supplementary Figure 1j–l). By contrast, we observed that—while anti-PD-L1 treatment had no effect on tumor growth and survival of mice over the first cycle of therapy—tumors started to regress during the second cycle, leading to significantly prolonged survival when compared to both NaCl 0.9% and cIgG controls (Figure 2g–i, Supplementary Figure 1m–o). Together, these findings suggest that immune checkpoint inhibition may have anti-tumor effects in high-risk neuroblastoma; however, these effects may occur only after a lag phase of treatment.

Combination of 6-thio-dG and anti-PD-L1 treatment overcomes the initial lag period in treatment response to anti-PD-L1 monotherapy

Next, we evaluated the treatment effects of anti-PD-L1 monotherapy versus the combination of 6-thio-dG and anti-PD-L1 treatment (Figure 2j–l). While we did not observe significant differences in survival of these two groups (p = 0.959), there was a remarkable delay in death of the disease in the group treated with combination therapy that was compensated after about 7 w of treatment (10 w of age; Figure 2j). In line with this observation, we noted that there was a slight benefit on tumor growth in the combination treatment group during the first weeks of treatment that was balanced out at later stages (Figure 2k). Together, these findings suggest that combining 6-thio-dG with anti-PD-L1 may improve tumor control by mitigating the lag phase of anti-PD-L1 monotherapy.

Combination of 6-thio-dG and anti-PD-L1 treatment induces cGAS and PD-L1 expression and immune cell infiltration in murine neuroblastoma

To explore the molecular mechanisms underlying the observed therapeutic effects, we performed immunoblot analyses for cGAS, PD-L1, IRF-3, and phospho-IRF-3 in tumor tissues obtained from mice of the distinct treatment groups (Figure 3a). Upregulation of cGAS was observed in 6-thio-dG treated tumors, both upon monotherapy and in combination with anti-PD-L1. This finding is consistent with the putative induction of telomere damage and subsequent activation of the cGAS-STING pathway by 6-thio-dG, as reported previously.¹⁶ In addition, PD-L1 expression was induced in mice treated with 6-thio-dG combined with either anti-PD-L1 or isotype control. By contrast, IRF-3 phosphorylation, a marker of STING pathway activation, was not detectable in murine tumors.

Figure 3.

Effects of 6-thio-dG and anti-PD-L1 antibody treatment on the cGAS-STING pathway and PD-L1 levels as well as immune cell infiltration in neuroblastomas of Th-MYCN;Th-ALK^F1174L mice. (a) cGAS, PD-L1, pIRF-3, and IRF-3 Immunoblot analysis of murine tumors of the distinct treatment groups (n = 2 per group). Treatment of NHO2A cells with the STING agonist DMXXA (20 µM for 24 h) served as a positive control; HSP-90 served as loading control. (b) HE staining and immunohistochemical staining for CD45, CD4, CD8, CD19, and CD11b of tumors of the distinct treatment groups. Images of one representative tumor out of ten tumors analyzed per group are shown. Arrows highlight cells positive for the respective marker. (c) Quantification of immune cells in treated and control tumors. Cells from five fields of view from one tumor per treatment group were counted and compared by one-sided Wilcoxon rank-sum test. Benjamini–Hochberg correction was performed to adjust for multiple testing. Details of comparisons are given in Supplementary Table 1.

Immunohistochemical staining of mouse tumors (n = 10 for each treatment group) revealed massive infiltration of immune cells in tumors treated with the combination of 6-thio-dG and anti-PD-L1 antibody (Figure 3b). Tumors from this group contained significantly more CD45+ leukocytes, CD8+ cytotoxic T cells, CD19+ B cells, and CD11b+ macrophages compared with those from control mice treated with NaCl 0.9% (Figure 3c). The combination also induced significantly higher infiltration of immune cells compared with 6-thio-dG monotherapy (CD45+, CD8+, CD19+, CD4+, and CD11b+) and 6-thio-dG plus cIgG (CD8+, CD19+, and CD11b+) (Figure 3b, c, Supplementary Table 1). Similarly, anti-PD-L1 monotherapy significantly increased CD45+, CD8+, CD4+, and CD11b+ immune cell infiltration relative to mice treated with NaCl 0.9% (Figure 3b, c, Supplementary Table 1). In contrast, tumors from mice treated with 6-thio-dG monotherapy exhibited overall lower immune cell infiltration, although CD8+ T cell numbers remained elevated when compared to tumors from NaCl 0.9% control mice (Figure 3b, c, Supplementary Table 1). Control Tumors (NaCl 0.9%, cIgG) contained only sparse immune cell populations (Figure 3b, c, Supplementary Table 1). These results indicate that 6-thio-dG combined with PD-L1 blockade effectively recruits immune cells to the tumor microenvironment.

Evidence for cGAS-STING pathway activity in human neuroblastoma

In human neuroblastoma cell lines, absence of baseline cGAS expression and lack of cGAS induction upon stimulation by cytosolic DNA have been reported,⁴⁰ whereas the functionality of the cGAS-STING pathway in primary neuroblastoma has remained uncertain. To assess whether the cGAS-STING pathway may be active in human neuroblastoma, we analyzed CGAS expression in primary tumor samples and compared it with that in human neuroblastoma cell lines (Figure 4a, Supplementary Table 2). Consistent with previous studies,⁴⁰ CGAS RNA expression was low in cell lines, but significantly higher in primary tumors. In line with this observation, we also detected cGAS protein in primary tumors as well as phosphorylation of the STING downstream effector IRF-3 (Figure 4b).

Figure 4.

Evidence for cGAS-STING pathway activity in human neuroblastoma. (a) CGAS expression in human neuroblastoma cell lines (n = 14) compared with primary tumor samples (n = 498, SEQC cohort). (b) cGAS, p-IRF-3, and IRF-3 Immunoblot analysis of THP-1 cells without treatment (positive control for cGAS protein expression), treated with diABZI (positive control for detection of pIRF-3), and five primary high-risk neuroblastoma tumors. (c) CGAS, STING1, IRF3, and CD274 (PD-L1) expression in high-risk neuroblastoma samples at diagnosis (n = 176) and in high-risk neuroblastoma samples under cytotoxic treatment (n = 10).

To assess whether cGAS-STING can be induced in human neuroblastoma, we determined transcript levels of CGAS, TMEM173, also known as STING1 and IRF-3 in primary tumors prior to and after exposure to cisplatin, a cytotoxic drug being used in neuroblastoma first-line treatment. We noted significantly increased expression of these genes in tumor samples obtained under cytotoxic treatment (Figure 4c, Supplementary Table 2). Furthermore, we also found increased CD274 (PD-L1) mRNA expression upon therapy (Figure 4c, Supplementary Table 2). Together, these findings suggest that the cGAS-STING pathway may be functional in at least a fraction of primary neuroblastomas, providing a basis for potential transfer of a telomerase-targeting concept into clinical practice.

Discussion

In this study, we demonstrate that targeting telomere integrity primes neuroblastoma for response to immune checkpoint inhibitors. Using an in vitro model of murine neuroblastoma cells and an in vivo transgenic mouse model of high-risk neuroblastoma, we showed that targeting telomeres with 6-thio-dG activates the cGAS-STING pathway, induces PD-L1 expression, and creates an immunologically active tumor microenvironment. In the mouse model, PD-L1 monotherapy delayed tumor growth and provided a survival benefit after an initial lag period in treatment response. Combining 6-thio-dG with anti-PD-L1 overcame this lag period and improved survival, promoted tumor regression, and enhanced immune cell infiltration. These findings support the concept of combining telomere dysfunction-inducing compounds with immune checkpoint blockade as a therapeutic strategy for high-risk neuroblastoma.

Previous studies have demonstrated that treatment with 6-thio-dG induces telomere damage in tumor cells that is recognized by dendritic cells, leading to activation of the cytosolic DNA-sensing STING pathway in the latter.¹⁶ This activation enhances the cross-priming capacity of dendritic cells and promotes activation of tumor-specific cytotoxic CD8+ T cells.¹⁶ Furthermore, direct STING activation by STING-activating nanoparticles enhanced anti-tumor immunity in neuroblastoma by inducing type I interferon signaling.³⁷ Our findings provide evidence for induction of the cGAS-STING pathway in neuroblastoma cells upon 6-thio-dG treatment, which may prime tumors for an anti-tumor innate immune response. In line with this notion, our in vivo experiments revealed increased immune cell infiltration of tumors after treatment with 6-thio-dG. A potential limitation of our study is the uncertainty regarding the functionality of the cGAS-STING pathway in human neuroblastoma. While this pathway is inactive in human neuroblastoma cell lines,⁴⁰ we noted significantly increased expression of CGAS in primary tumors over cell lines, in general, and a further increase of CGAS and its downstream effector genes STING1 and IRF-3 in tumor samples obtained under cytotoxic treatment, which is in line with activation of the cGAS-STING pathway by cisplatin, as observed in other cancer types.^41,42 These results provide evidence for its functionality in primary human tumors, thus supporting the concept of cGAS-STING induction by 6-thio-dG.

We also observed that treatment of Th-MYCN; Th-ALK^F1174L transgenic mice with PD-L1 monotherapy resulted in prolonged overall survival and tumor regression. These findings contrast with previous studies investigating the efficacy of anti-PD-L1 therapy in the transgenic Th-MYCN mouse model, which reported no therapeutic effects.^24,25 A possible explanation for this discrepancy is the early experimental endpoint in prior studies, where mice were sacrificed 10 d after treatment initiation. In this study, tumor regression was observed only after four weeks of treatment, suggesting a lag period before treatment becomes effective. Our findings suggest that high-risk neuroblastoma patients may benefit clinically from checkpoint inhibition therapies, but that the initial lag phase in treatment response poses a challenge that requires additional therapy. The delay in treatment response may also explain why anti-PD-L1/anti-PD-1 monotherapy has shown limited efficacy in clinical studies on neuroblastoma.^29,30

Treatment with 6-thio-dG led to upregulation of PD-L1 in vitro, a phenomenon reversed by STING antagonist SN-011 and mimicked by STING agonist DMXXA, which suggests a mechanistic link between telomere damage, cGAS-STING pathway activation, and PD-L1 expression,^37,43,44 thereby suppressing T cell function by engaging PD-1 receptors on cytotoxic T cells.⁴⁵ This finding, however, also suggests that 6-thio-dG treatment may create a vulnerability for immune checkpoint blockade, as reported previously.¹⁶ Based on these results, we hypothesized that 6-thio-dG may prime neuroblastoma for immune checkpoint blockade therapy in vivo, and that this strategy may overcome the lag phase observed in anti-PD-L1 monotherapy. We indeed found that the combination of 6-thio-dG and anti-PD-L1 therapy delayed tumor growth and extended survival in Th-MYCN;Th-ALK^F1174L transgenic mice compared to 6-thio-dG monotherapy. We also found a higher fraction of partially regressing tumors in mice treated with both 6-thio-dG and anti-PD-L1 treatment. Together, these findings suggest that 6-thio-dG induced PD-L1 expression may sensitize tumors to immune checkpoint blockade therapy in high-risk neuroblastoma, as it has been reported in other cancer types.¹⁶

Survival of mice treated with a combination of 6-thio-dG with IgG control antibody did not differ from either mice treated with 6-thio-dG monotherapy or in combination with anti-PD-L1 antibody, although a higher fraction of tumor regression was observed in the latter group. We also noticed massive immune cell infiltration in tumors treated with 6-thio-dG plus IgG control, similar to those treated with 6-thio-dG plus anti-PD-L1 antibody, suggesting a non-specific immune response to the xenogeneic rat IgG used in this experiment. Together, the precise impact of control IgG in combination with 6-thio-dG in our study has remained inconclusive and may need further evaluation.

In summary, our study demonstrates that 6-thio-dG treatment can activate the cGAS-STING pathway and induce PD-L1 expression in neuroblastoma. Combining 6-thio-dG with anti-PD-L1 therapy is associated with intratumoral immune cell accumulation, tumor regression, and improved survival in an aggressive murine model of high-risk neuroblastoma. Together, these findings provide a rationale to further investigate telomere dysfunction and immune checkpoint blockade as a combined therapeutic strategy in high-risk neuroblastoma.

Supplementary Material

Supplementary material

Supplementary file

Supplementary Table 1.xlsx

Supplementary Table 1.xlsx

Supplementary Table 2.xlsx

Supplementary Table 2.xlsx

Supplementary_figure_1 (1).tif

Supplementary_figure_1 (1).tif

Funding Statement

This work was funded by the Deutsche Krebshilfe through a Mildred Scheel Nachwuchszentrum Grant (Grant number 70113307 to SH and JB). This work was supported by the German Research Foundation (DFG, Deutsche Forschungsgemeinschaft) as part of CRC1399 (grant ID 413326622 to LW, JB, MH, and MF), CRC1588 (grant ID 493872418 to MF), and CRC1310 (grant ID 325931972 to JB). It was also supported by the CANTAR network (NW21-062B to JB, MH, and MF), an initiative of the Ministry of Science of the State Northrhine Westphalia, Germany. MF received funding from Leverkusen hilft krebskranken Kindern.

Disclosure of potential conflicts of interest

SH, LW, BS, CB, AMH, CR, EL, GP, YK, NI, NH, WL, MH, and MF have no financial interests. JB has received funding for research from Bayer and for travel from Merck outside of the submitted work.

Data availability statement

The data that support the findings of this study are available from the corresponding author (MF), upon reasonable request.

Ethics statement

The clinical trials were approved by the institutional ethical review board of the University of Cologne (votes 9764, 04-049, and 16-432). The use of patient tumor material for research was approved by the institutional ethical review board of the University of Cologne (vote 23-1387). All mouse experiments were conducted in accordance with FELASA recommendations, the ARRIVE guidelines, and were approved by the local animal care committee and relevant authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, AZ: 81-02.04.2021.A395).

Disclaimers

The content of this article is expressing our own view and is not an official position of the institution or funders.

Supplemental material

Supplemental data for this article can be accessed at https://doi.org/10.1080/2162402X.2026.2653918.