RRM1 inhibition sensitizes lung adenocarcinoma to decitabine treatment

TL;DR

Blocking RRM1 makes lung cancer cells more responsive to decitabine treatment by increasing drug incorporation and activating tumor suppressor genes.

Contribution

The study identifies RRM1 inhibition as a novel strategy to enhance decitabine efficacy in lung adenocarcinoma.

Findings

RRM1 levels inversely correlate with decitabine incorporation in lung cancer cells.

RRM1 inhibition increases decitabine's effect on tumor suppression and STING pathway activation.

Reducing RNR activity boosts decitabine incorporation by lowering dCTP availability.

Abstract

Aberrant DNA methylation has been implicated in tumorigenesis and the development of lung cancer. However, Nucleoside analog DNA methyltransferase inhibitors have demonstrated clinical utility in the treatment of myelodysplastic syndrome and acute myeloid leukemia; the drugs have not shown commensurate clinical efficacy in solid tumors. Mechanisms mediating the primary resistance to DNA hypomethylating agents in solid tumors are not fully understood. Here, we hypothesized that factors that limit incorporation of nucleoside analog DNA methyltransferase inhibitors in genomic DNA may underlie the tumor cell intrinsic primary resistance to decitabine (DAC) in lung cancer. We found that RRM1 expression levels were inversely correlated with DAC incorporation rates detected by LC–MS/MS. RNA interference-mediated depletion of RRM1, the catalytic subunit of ribonucleotide reductase (RNR), or pharmacological inhibition of RNR significantly potentiated inhibition of lung cancer cell clonogenic survival in vitro and xenograft growth in vivo by DAC treatment. Additionally, RRM1 inhibition enhances DAC-mediated tumor suppressor gene reactivation and STING pathway activation via DNA damage-induced IFI16 sensing. RNR inhibition led to increased incorporation of DAC into genomic DNA by reducing the availability of dCTP. These findings nominate the promising combination therapy of DAC and RNR inhibitors as being ripe for further clinical translation.

Figures

RRM1 level correlates with the incorporation rate of DAC into genomic DNA.A and B Dose-responsive effect of DAC incorporation and 5mC loss in response to DAC exposure. The LUAD cell line A549 was treated with a concentration series of DAC (0.5, 1, 2, 5, 10 μM) for 72 h, and the contents of DAC (A) and 5mC (B) in genomic DNA were detected by LC–MS/MS as a ratio to the content of dC. C Schematic representation of the screen to identify the cell-intrinsic factors that may be involved in limiting DAC incorporation in tumor cells. D The IC_50_ of DAC in mediating cytotoxicity as measured by Alamar

Inhibition of RRM1 sensitizes lung adenocarcinoma cells to DAC in vitro and in vivo.A RRM1 mRNA expression level in lung adenocarcinomas and squamous cell carcinomas is significantly higher than that in normal lung tissues. RRM1 expression data were downloaded from Xena Functional Genomics Explorer ([https://xena.ucsc.edu/](https://xena.ucsc.edu/)) on 12/5/2024, including The Cancer Genome Atlas (TCGA), Therapeutically Applicable Research to Generate Effective Treatments (TARGET), and Genotype-Tissue Expression (GTEx). The *Z*-score of RRM1 expression in each cancer sample was normalized to av

Pharmacological inhibition of RNR cooperates with DAC treatment to suppress lung adenocarcinoma cell growth in vitro and in vivo.A A549, PC-9, and HOP62 cell lines were seeded into 12-well plates at the rate of 200 cells/ well and treated with 200 nM 3-AP and 50 nM DAC as single agents or in combination. The colony formation ability of A549, PC-9 and HOP62 was shown (*n* = 3). B and C Female BLAB/c nude mice aged 4–6 weeks were taken in the study. 2 × 10^6^ PC-9-wild type (WT) cells were inoculated subcutaneously. When the tumor volume reached 100–150 mm^3^, mice were randomly divided into 4 g

RRM1 knockdown and inhibition increase DAC effects on DNMT1 level and biomarker gene expression*.*A The LUAD cell lines A549 and PC-9 with or without RRM1 knockdown were treated with DMSO or DAC (100 nM) for 6 days. Every 48 h change the medium and drug. Protein samples were extracted, and a western blot was used to detect the expression levels of DNMT1 and RRM1. GAPDH serves as the loading control (*n* = 3). B The LUAD cell lines A549 and PC-9 were treated with DMSO, DAC (50, 100 nM), and 3-AP (200 nM) for 6 days. Every 48 h change the medium and drug. Protein samples were extracted, and a we

RRM1 inhibition promotes DAC-induced global epigenetic/reprogramming effects and the immune response pathway.A In the RNA-seq experiment, four groups were divided: Control group (Ctrl), DAC treatment group (DAC), RRM1 knock-down group (RRM1-KD), and DAC plus RRM1 knock-down group (COMB). The Volcano plots showed differentially expressed genes between the Comb (unique) group compared to the DAC group. Red: significantly up-regulated genes. Blue: significantly down-regulated genes. B The gene ontology (GO) enrichment analysis across comparison groups: COMB (unique) versus control. C and D A549 w

RRM1 modulation has differential effects on different nucleoside analog DNMT inhibitors (AZA and DAC).A Inhibition of growth and viability of human lung cancer cell lines HOP62 with DAC and 3-AP. HOP62 cells were treated with the indicated dose range of DAC and 3-AP in the Incucyte time-lapse system. Growth curves were measured as the percent of total confluence in each well over time. The mean percent confluence ± SEM for 16 fields in each condition is shown, with measurements taken every 6 h throughout the course of the growth curve. Viability measurements of cells treated with DAC and 3-AP

Full text

Introduction

Lung cancer is one of the most common and lethal cancer types [1], comprised of non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC). Although surgical treatment is effective in the early stages, the 5-year overall survival rate of NSCLC is still only about 15%, because most patients have been diagnosed in the advanced stages [2]. Targeted therapy with tyrosine kinase inhibitors (TKI) has achieved encouraging efficacy, but the limited prevalence of targetable driver mutations and primary and secondary drug resistance have remained major challenges. Further, immune checkpoint inhibition is only effective in a small fraction of patients [3]. Thus, new therapeutic approaches are needed to expand the clinical benefit to a wider range of NSCLC patients.

Epigenetic dysregulation is nearly universal in human cancers. Particularly, abnormal DNA methylation, catalyzed by DNA methyltransferases (DNMTs), is considered an important player in the occurrence and development of various cancer types, including NSCLC [4]. Among multiple mechanisms, DNA hypermethylation at promoter-associated CpG islands transcriptionally silences tumor suppressor genes [5]. Considering the reversibility of DNA methylation, DNA hypomethylating agents, particularly DNMT inhibitors, have attracted much attention in developing new cancer treatments. Decitabine (DAC) and azacitidine (AZA) are nucleoside analog DNMT inhibitors that are effective in the treatment of myelodysplastic syndrome (MDS) and acute myelocytic leukemia (AML) [6]. However, such nucleoside analog DNMT inhibitors have not yet demonstrated success in solid tumors [7, 8]. Specifically, single-drug treatment with DAC in NSCLC is ineffective [9], underscoring the necessity of uncovering the primary resistance mechanisms.

DAC is taken up by cancer cells, metabolized to DAC-triphosphate, and then incorporated into genomic DNA during DNA synthesis, leading to trapping of DNMT enzymes and passive demethylation [10]. Factors that limit the incorporation of DAC into genomic DNA can lead to primary or acquired resistance to DAC. The nucleotidases dCTP pyrophosphatase 1 (DCTPP1) and sterile alpha motif and histidine-aspartate domain-containing protein 1 (SAMHD1), and the deaminases cytidine deaminase (CDA) and dCMP deaminase (DCTD) have previously been shown to limit incorporation of DAC into genomic DNA, leading to resistance [11–15]. Specifically, DCTPP1 functions as a deoxycytidine triphosphate pyrophosphatase that hydrolyzes the pyrophosphate group of active metabolite 5-aza-dCTP, reducing its availability for genomic incorporation [11]. SAMHD1, a deoxynucleoside triphosphohydrolase activated by dGTP, cleaves 5-aza-dCTP into inactive metabolites, lowering intracellular levels of the active drug and impeding its DNA integration [12, 15]. CDA mediates rapid deamination of DAC to form inactive 5-aza-deoxyuridine, accelerating drug clearance and reducing the effective concentration available for DNA incorporation [13, 14]. DCTD catalyzes the deamination of DAC monophosphate (5-aza-dCMP) to 5-aza-deoxyuridine monophosphate, diminishing the pool of precursors required for 5-aza-dCTP synthesis and thereby limiting genomic incorporation of DAC [13]. Here, we identify RRM1, the catalytic subunit of ribonucleotide reductase (RNR) [16, 17], as an important factor in limiting the incorporation rate of DAC into genomic DNA. We show that genetic knockdown of RRM1, or pharmacological inhibition of RNR, can strongly sensitize cancer cells to DAC. Mechanistically, inhibition of RNR may reduce the entry of cytosine nucleoside into the dNTP pool, increase the utilization of DAC for genomic DNA synthesis, thus leading to a higher incorporation rate. The synergistic combination of DAC with RNR inhibitors is therefore ripe for further preclinical and clinical translation.

Materials and methods

Cell culture

The HOP62, BT549, DU145, NCI-H23, NCI-H322M, NCI-H460, HOP92, MCF7, PC3, OVCAR8, SF295, T47D, MDA-MB-231, MDA-MB-435, and SNB75 cell lines were obtained from American Type Culture Collection (ATCC). The A549, PC-9, and HEK293T cell lines were purchased from the Chinese Collection of Authenticated Cell Cultures (Shanghai, China). The murine Kras^G12C^/Trp53^−/−^ (KP) lung cancer cell line was a gift from Professor Fei Li at Fudan University [18]. The BT549, H23, H322M, HOP92, OVCAR8, SF295, and T47D cell lines were cultured in RPMI1640 medium (BasalMedia) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. The BDM154, H460, MCF7, PC-3, MDA-MB-231, MDA-MB-435, SNB75, A549, PC-9, KP, HOP62, and HEK293T cell lines were cultured in high-glucose DMEM medium (BasalMedia) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. All cell line cultures were maintained under standard culture conditions (37 °C, 5% CO_2_). The cell lines were tested for Mycoplasma at regular intervals.

Chemicals

Decitabine (DAC, 5-Aza-2‘-deoxycytidine, #HY-A0004, CAS No. 2353-33-5), triapine (3-AP, #HY-10082, CAS No. 143621-35-6), clofarabine (CLO, #HY-A0005, CAS No. 123318-82-1), and hydroxyurea (HU, #HY-B0313, CAS No. 127-07-1) were purchased from MedChemExpress (MCE) chemicals.

RNA interference

Human-RRM1 shRNA (target sequence: GGACAAGACCAGCGGCUAATT) and control shRNA were synthesized and verified by sequencing by Beijing Tsingke Biotech Co., Ltd. The shRNA was cloned into pLKO.1 puro (Addgene #8453). The lentiviral particles were generated by transfecting shCtrl or shRRM1, pMD2.G (Addgene #12259), and psPAX2 (Addgene #12260) packaging plasmids together with transfer plasmid into HEK293T cells using HighGene transfection reagent (ABclonal # RM09014) following the manufacturer’s instructions.

CRISPR/Cas9 system

Human STING-targeted sgRNA sequences (GCAGGCACTCAGCAGAACCA) and mouse Sting-targeted sgRNA sequences (TATCTCGGAATCGAATGTTG) were designed using the sgRNA designer (https://portals.broadinstitute.org/gppx/crispick/public). A non-targeting sgRNA from the Gecko library v2 was used as a scramble sgRNA. sgRNAs were cloned into lentiCRISPR v2 (Addgene #52961). Lentiviral particles were prepared by transfecting HEK293 cells with sgScramble or sgSTING, pMD2.G (Addgene #12259), and psPAX2 (Addgene #12260) using HighGene transfection reagent (ABclonal # RM09014). Stable cell lines were generated, followed by puromycin selection.

Cell viability assay

Cell viability was determined by using Cell Counting Kit-8 (Yeasen #40203ES76). Tumor cells in DMEM medium supplemented with 10% FBS were plated in 96-well plates (2000 cells/100 μL/well). After 24 h of culture, the complete medium containing different concentrations of dissolved drugs (50 nM DAC and 100 nM 3-AP) was added to the plates. The blank control group was treated with 0.1% DMSO vehicle. More than 3 multiple wells were set up in each group, and incubation time points of 0, 24, 48, 72, and 96 h were set, respectively. After cultivation, 10%(volume/volume) of CCK-8 reagent was added to each well according to the instructions. The 96-well plates were incubated at 37 °C for 1 h, and the absorbance at 450 nm wavelength was measured by SpectraMax iD5 (Molecular Devices). The relative cell viability compared to the vehicle control was determined based on absorption values.

Time-lapse Incucyte growth assay

500 cancer cells were seeded per well in 48-well plates one day before compound treatments. DAC, 3-AP, and other drugs were freshly prepared and added to culture plates at the indicated concentrations. For growth curve experiments, growth data were determined by imaging cell confluence at 6 h intervals at each of 16 fields per well using the IncuCyte™ ZOOM System (Essen Bioscience, MA) in the Johns Hopkins SKCCC Imaging Core Facility. The percentage confluence was calculated with Incucyte built-in software. Growth curves were analyzed and constructed with Graphpad Prism software 9.0 (La Jolla, CA).

Colony formation assay

Tumor cells were seeded in 12-well plates at a density of 200 cells per well. The cells were treated with the corresponding compounds after 24 h. The medium was replaced every 48 h, with freshly added compounds. After 10–12 days of treatment, cell colonies were fixed with 4% formaldehyde and stained with 0.25% crystal violet. Pictures of stained cells were taken using a digital camera. Area of cell colonies were quantified by Image J software (National Institutes of Health, USA). For the quantification of colony formation efficiency, normalization was performed to explicitly evaluate RRM1-mediated sensitization to DAC: raw colony formation ratios were first calculated for each group by dividing the number of formed colonies by the initial plated cell number (200 cells per well), followed by normalization of these raw ratios to the mean value of the control group (designated as 100%) to derive the relative colony formation efficiency.

Quantitative real-time PCR and RT-PCR

RNA was extracted by total RNA isolation reagent (Biosharp #BS258A) according to the manufacturer’s instructions. Reverse transcription (RT) was performed using 1 μg of total RNA and NovoScript® 1st Strand cDNA Synthesis SuperMix kit (Novoprotein #E043-01A). Quantitative PCR was performed using NovoStart® SYBR qPCR SuperMix plus (Novoprotein, #E096-01A) on LightCycler480II (Roche) according to the manufacturer’s protocol. The specificity of the PCR amplification was validated by agarose gel electrophoresis and melting curve analyses. Values represent the average of three technical replicates from at least three independent experiments (biological replicates), normalized by using the GAPDH gene as an internal control. qPCR results were quantified using the ΔΔCt method. Sequences of PCR primers are listed in Supplementary Table 1.

Western blot

Tumor cells were lysed in RIPA buffer (Beyotime #P0013C) supplemented with protease and phosphatase inhibitor cocktail for general use (Beyotime #P1045). The 20 μg of protein was loaded onto 10% or 12% SDS–polyacrylamide gels, separated by electrophoresis and electrotransferred onto a nitrocellulose filter membrane (Millipore). Membranes were blocked with 5% skim milk in Tris-buffered saline with Tween-20 (TBST) for 1 h at room temperature, followed by overnight incubation with primary antibodies at 4 °C. After washing with TBST, membranes were incubated with appropriate secondary antibodies at room temperature for 1 h. Antibody–antigen complexes were detected using ECL Western Blotting Substrate (Yeasen #36222ES60). Antibody information is listed in Supplementary Table 2.

DNA dot blot analysis

Genomic DNA was extracted from cells using a TIANamp Genomic DNA Kit (TIANGEN #DP304). Denatured 500 ng DNA was spotted on a nitrocellulose membrane (Waterman) and crosslinked by ultraviolet irradiation. Briefly, the membrane was first blocked with 5% milk in TBST for 1 h and then incubated with an anti-5-methylcytosine antibody (1:2500, Proteintech #68301) overnight at 4 °C. Following this, the membrane was incubated with appropriate secondary antibodies for 1 h at room temperature, washed three times with TBST, and finally the DNA was visualized using ECL Western Blotting Substrate (Yeasen #36222ES60).

Flow cytometry analysis

For cell cycle analysis, a total of 1 × 10^6^ cells were fixed with precooled 70% ethanol overnight and then processed using the Cell Cycle and Apoptosis Analysis Kit (Yeasen #40301ES50) according to the manufacturer’s instructions. The flow cytometry data were generated on a Cytoflex LX (Beckman-Coulter) flow cytometer and analyzed with FlowJo v10 software.

Terminal deoxynucleotidyl transferase-mediated Nick end labeling (TUNEL)

To detect apoptosis, cell climbing slides were first placed in 24-well plates, followed by seeding 5 × 10^4^ A549 shCtrl or shRRM1 cells per well. The next day, the medium was replaced with fresh medium containing 100 nM DAC, and the cells were cultured for 6 days. Subsequently, cells were fixed with 4% paraformaldehyde at 4 °C for 25 min, then permeabilized with 0.2% Triton X-100 prepared in phosphate-buffered saline (PBS) for 5 min at room temperature. Afterward, 50 μL of terminal deoxynucleotidyl transferase (TdT) incubation buffer (Yeasen #40306ES) was added to each well, and the cells were incubated at 37 °C for 60 min. Finally, the cells were stained with 1 μg/mL propidium iodide (PI) solution and 2 μg/mL 4’,6-diamidino-2-phenylindole (DAPI) solution sequentially, and observed immediately under a fluorescence microscope.

Immunohistochemistry staining

Tissues were fixed in 4% formalin overnight and embedded in paraffin. Tumor blocks were cut into 4 μm-thick sections. Slides were treated with 3% hydrogen peroxide for 15 min at room temperature to block endogenous peroxidase activity. Antigen was retrieved by microwaving the tissue sections in 10 mM citrate buffer (pH 6.0). Primary antibody was applied and incubated at 4 °C overnight. The DAB Horseradish Peroxidase Color Development Kit (Beyotime #P0202) was used to detect the immunoreactivity according to the manufacturer’s manual. The following primary antibodies were used: Ki67 (1:500 dilution; Servicebio #GB111141). All samples were stained with hematoxylin-eosin.

Quantitation of genomic DAC content by high-pressure liquid chromatography-tandem mass spectrometry (LC–MS/MS)

The DAC content in genomic DNA was determined by a high-pressure liquid chromatography/tandem mass spectrometry (LC–MS/MS) procedure as described previously [19]. Briefly, 1 µg genomic DNA was digested with 4 Units of Nuclease P1 (Sigma #N8630) at 65 °C for 10 min. Subsequently, the DNA was dephosphorylated by using 4 Units of Alkaline Phosphatase (Roche Life Science #APMB-RO) at 37 °C for 1 h. Standards and quality control (QCs) samples were prepared by adding known concentrations of DAC, 2′-deoxycytidine (dC), and 5-methyl-2′-deoxycytidine (5mC) into a blank digest matrix. The 5-azacitidine(5AC)-^15^N_4_ 2dC-^13^C^15^N_2_, and 5mC-d_3_ were used to be the internal standard. The samples were injected onto a LC–MS/MS system consisting of a Waters Acquity UPLC (Milford, MA) interfaced with an AB Sciex 5500 triple quadruple mass spectrometer (Foster City, CA). Chromatographic separation was achieved using a Thermo Hypercarb porous graphite analytical column (100 × 2.1 mm, 5 μm, Waltham, MA) running isocratic elution with a mobile phase consisting of 10 mM ammonium acetate: acetonitrile with 0.1% formic acid (70:30, v/v) at a flow rate of 0.3 μL/mL. The mass spectrometer was run in positive electrospray ionization mode, monitoring for the following MRM transitions: 5-aza-dC (DAC): 228.9 → 113.0, 2dC: 228.0 → 112.0, 5-methyl-dC: 242.0 → 126.0, 5AC-15N4 (internal standard): 249.0 → 117.0, 230.8 → 115.0 for 2dC-13C15N2 and 245.8 → 129.0 for 5mC-d3. The calibration range was 2–400 ng/mL for 5-aza-dC, 5–1000 ng/mL for 5-methyl-dC, and 50–10,000 ng/mL for 2dC. All analytes used quadratic regression with 1/x2 weighing. Results were reported as DAC per thousand dC.

Growth inhibition of subcutaneous xenografted tumors

Sample size was determined based on commonly used sample sizes in similar studies within the field, with 5 mice assigned to each group to ensure the stability and reliability of the experimental results. Female BALB/c-nude mice (5–6 weeks) were purchased from Shanghai Experimental Animal Center (Chinese Academy of Sciences, Shanghai, China). 3 × 10^6^ suspended PC-9 shRRM1 or PC-9 shCtrl cells (in 100 μL PBS) were injected into the right posterior flank of each mouse. When the tumors reached ~50–100 mm^3^ in size, mice were randomly divided into each treatment arm and received intraperitoneal injection of appropriate drug as outlined in Fig. 2C: shCtrl (PBS), shCtrl (DAC, three 0.5 mg/kg/doses given i.p. every other day on week 1–3), shRRM1 (PBS), shRRM1 (DAC, three 0.5 mg/kg/doses given i.p. every other day on week 1–3). Tumors were measured every 3 days using a vernier caliper, and the tumor volume was calculated by the following formula: V = (Length*Width^2^)/2. At the endpoint, mice were sacrificed, and the tumors were collected for further analysis. No animals or samples were excluded from the analysis unless pre-defined quality control criteria were not met. The experiments were performed according to the regulations for the administration of affairs concerning experimental animals and were approved by the experimental animal ethics committee of Zhejiang University.

In vivo combination of DNMTi and RNRi assay

Sample size was determined based on commonly used sample sizes in similar studies within the field, with 5 mice assigned to each group to ensure the stability and reliability of the experimental results. Female BALB/c-nude mice (5–6 weeks) were purchased from Shanghai Experimental Animal Center (Chinese Academy of Sciences, Shanghai, China). 3 × 10^6^ suspended PC-9 cells (in 100 μL PBS) were injected into the right posterior flank of each mouse. When the tumors reached ~50–100 mm^3^ in size, mice were randomly divided into each treatment arm and received intraperitoneal injection of appropriate drug as outlined in Fig. 3B: control (PBS), DAC (three 0.5 mg/kg/doses given i.p. every other day on week 1–3), 3-AP (20 mg/kg/doses given i.p. each day for 5 days with 2 days’ rest each week), or both. Male C57BL/6 (5–6 weeks) were purchased from Shanghai Experimental Animal Center (Chinese Academy of Sciences, Shanghai, China). 0.5 × 10^6^ suspended KP cells (in 100 μL PBS) were injected into the right posterior flank of each mouse. When the tumors reached ~50–100 mm^3^ in size, mice were randomly divided into each treatment arm and received intraperitoneal injection of appropriate drug as outlined in Fig. 5I: control (PBS), DAC (0.1 mg/kg/doses given i.p.), 3-AP (20 mg/kg/doses given i.p.), or both. Tumors were measured every 2 or 3 days using a vernier caliper, and the tumor volume was calculated by the following formula: V = (Length*Width^2^)/2. At the endpoint, mice were sacrificed, and the tumors were collected for further analysis. No animals or samples were excluded from the analysis unless pre-defined quality control criteria were not met.

RNA-Sequencing and data analysis

1 μg total RNA was used for the following library preparation. The RNA-seq libraries were constructed using the VAHTS Universal V8 RNA-seq Library Prep Kit for Illumina (Vazyme #NR605-02). The poly(A) mRNA isolation was performed using Oligo(dT) beads. The mRNA fragmentation was performed using divalent cations and high temperature. Priming was performed using Random Primers. First-strand cDNA and the second-strand cDNA were synthesized. The purified double-stranded cDNA was then treated to repair both ends and add a dA-tailing in one reaction, followed by a T-A ligation to add adapters to both ends. Size selection of Adapter-ligated DNA was then performed using DNA Clean Beads. Each sample was then amplified by PCR using P5 and P7 primers, and the PCR products were validated. Then libraries with different indexes were multiplexed and loaded on an Illumina NovaSeq 6000 instrument for sequencing using a 2 × 150 paired-end (PE) configuration according to the manufacturer’s instructions. The analysis of differential gene expression was conducted with the use of the “DESeq2” package. The significantly expressed genes were defined as those genes with the value of log_2_ (fold change) > 1.5 and P-value < 0.05. Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis was performed by using the “clusterProfiler” package.

Blinding procedures

To minimize potential bias during data acquisition and analysis, several assessments were performed in a blinded manner. Specifically, TUNEL images used for apoptosis detection were randomized and analyzed by investigators blinded to group assignments. For Western blot and qPCR analyses, blinding was not applied, as data were quantified by using automated software based on band intensity or Ct values. Blinding was not applied in the animal experiment to ensure the consistency of operations throughout the study.

Statistics and plots

All experiments were performed at least three independent times, and consistent, reproducible results were obtained. Statistical differences were considered significant at p-value < 0.05. Statistical analysis was performed using Prism GraphPad v 9.0 software (San Diego, CA). For comparisons between sample pairs, two-tailed Student’s t-tests were used. For experiments with more than two conditions, one-way or two-way ANOVA followed by multiple comparisons tests were used.

Results

RRM1 level correlates with the incorporation rate of DAC into genomic DNA

To assess 5mC content and DAC incorporation into genomic DNA, we used liquid chromatography with tandem mass spectrometry (LC–MS/MS) as described previously (Supplemental Fig. S1A) [19]. As expected, we observed a strong positive correlation between DAC incorporation rates into genomic DNA and treatment doses (Fig. 1A). Correspondingly, 5mC content (measured as a ratio to dC) showed a dose-dependent reduction (Fig. 1B). Using this approach, we previously measured DAC incorporation in a panel of 16 tumor cell lines (5 lung cancer, 4 breast cancer, 4 melanoma, 2 prostate cancer and 1 ovarian cancer), and carried out a new analysis to assess the correlation with the half maximal inhibitory concentration of DAC (IC50) in the same cell lines (Fig. 1C) [11]. It was found that the higher the DAC integration rate, the lower the IC_50_ of cells to DAC, presenting a negative correlation (Fig. 1D). These results suggest that enhancing the incorporation of DAC into genomic DNA may promote the antitumor effects of DAC.Fig. 1RRM1 level correlates with the incorporation rate of DAC into genomic DNA.A and B Dose-responsive effect of DAC incorporation and 5mC loss in response to DAC exposure. The LUAD cell line A549 was treated with a concentration series of DAC (0.5, 1, 2, 5, 10 μM) for 72 h, and the contents of DAC (A) and 5mC (B) in genomic DNA were detected by LC–MS/MS as a ratio to the content of dC. C Schematic representation of the screen to identify the cell-intrinsic factors that may be involved in limiting DAC incorporation in tumor cells. D The IC_50_ of DAC in mediating cytotoxicity as measured by Alamar blue assay is inversely correlated with the rate of DAC incorporation into genomic DNA in 16 tumor cell lines from NCI-60 (4 breast cancers (BT549, MDA-MB-231, MCF7, T47D), 5 lung cancers (A549, HOP92, NCI-H23, NCI-H322, NCI-H460),1 ovarian cancer (OVCAR8), 4 melanomas (Malme-3M, UACC-257, UACC-62, MDA-MB-435), and 2 prostate cancers (DU145, PC3)). E A plot of the correlation coefficient versus the [–log10(p-value)] for the correlation between 5-aza-dC incorporation into genomic DNA and the level of mRNA expression for each gene from the Kyoto Encyclopedia of Genes and Genomes (KEGG) “ribonucleoside biosynthesis” and “deoxyribonucleoside biosynthesis” pathways. The Human Genome Organization (HUGO) gene symbols for the top-ranked genes (with p < 0.05) are indicated. F The correlation of DAC integration rates of 16 typical tumor cells from NCI-60 treated with DAC and the RRM1 expression levels of these 16 cell lines from the cBioportal database was conducted.

We thus sought to identify cell-intrinsic factors that may be involved in limiting DAC incorporation in tumor cells, since these factors may represent cell-intrinsic resistance mechanisms to DAC. We interrogated the expression of nucleoside metabolism genes in the above-mentioned 16 cell lines [20, 21], and analyzed their correlation with the measured DAC incorporation rates (Supplementary Fig. S1B, C). Interestingly, among genes in the “ribonucleoside and deoxyribonucleoside biosynthesis” pathway curated in KEGG, RRM1 showed the top inverse correlation with DAC incorporation rate (Fig. 1E, F and supplementary Fig. S1D). The degree of inverse association was similar to that observed for DCTPP1, which has previously been shown to be a cancer cell intrinsic resistance gene for DAC [11]. These data suggest that RRM1 may mediate cancer cell intrinsic resistance to DAC.

Knocking-down RRM1 sensitizes lung adenocarcinoma cells to DAC treatment in vitro and in vivo

We next investigated the impact of RRM1 inhibition on the antitumor effect of DAC treatment. By analyzing the public datasets from The Cancer Genome Atlas (TCGA), Therapeutically Applicable Research to Generate Effective Treatments (TARGET), and Genotype-Tissue Expression (GTEx), we found that RRM1 expression levels were significantly upregulated in lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) tissues compared to those in normal lung tissues (Fig. 2A); indeed, lung cancers had among the highest levels of RRM1 upregulation across all major cancer types (Supplementary Fig. S2A). LUAD patients with low RRM1 expression had a better prognosis (Supplementary Fig. S2B). In addition, RRM1 protein and RNA levels were overexpressed in two LUAD cell lines, A549 and PC-9, in comparison with human normal bronchial epithelial (HBE) cells (Supplementary Fig. S2C, D). We then treated A549 and PC-9 cells with different doses of DAC and showed that DAC treatments caused DNMT1 decrease and loss (Supplementary Fig. S2E, F), which is consistent with previous reports [22]. Notably, within cell-specific low concentration ranges (<500 nM for A549 cells and <200 nM for PC-9 cells), DAC reduced DNMT1 protein levels in a dose-dependent manner—with higher concentrations yielding more pronounced reductions. In contrast, high-dose DAC (exceeding these thresholds) was less effective at decreasing DNMT1. (Supplementary Fig. S2E, F); this may be attributable to the severe cell cycle arrest due to DNA damage from high-dose DAC, leading to dampening of DAC incorporation, as has been suggested previously [23, 24]. Accordingly, within the low-dose range of DAC, there was significant growth inhibition in A549 and PC9 cells as measured by cell colony formation assays (Supplementary Fig. S2G).Fig. 2. Inhibition of RRM1 sensitizes lung adenocarcinoma cells to DAC in vitro and in vivo.A RRM1 mRNA expression level in lung adenocarcinomas and squamous cell carcinomas is significantly higher than that in normal lung tissues. RRM1 expression data were downloaded from Xena Functional Genomics Explorer (https://xena.ucsc.edu/) on 12/5/2024, including The Cancer Genome Atlas (TCGA), Therapeutically Applicable Research to Generate Effective Treatments (TARGET), and Genotype-Tissue Expression (GTEx). The Z-score of RRM1 expression in each cancer sample was normalized to average expression levels in the respective normal tissues. p-values were calculated using a one-sided t-test. B A549 and PC-9 cell lines with or without RRM1 knockdown were seeded into 12-well plates at the rate of 200 cells/ well, and cell colony formation experiments were performed with 50 and 100 nM DAC treatment for 6 days (n = 3). C, D Female BALB/c nude mice aged 4–6 weeks were taken in the study. 2 × 10^6^ PC-9 cells with or without RRM1 knockdown were inoculated subcutaneously. When the tumor volume reached 100–150 mm^3^, mice were randomly divided into 4 groups to start the drug administration. Two weeks after drug administration, the tumor was removed and photographed (n = 5 per group). E and F The long diameter and short diameter of the tumor were measured every 2 days, and the tumor volume and tumor weight were calculated for statistical analysis (n = 5 per group). G The representative images of tumor tissues were obtained and fixed, and sliced for immunohistochemical detection of tumor proliferation marker Ki67. H The weight of mice was recorded every 2 days and analyzed statistically. I Blood samples were collected from the posterior orbital venous plexus before and after drug intervention, and blood routine, liver, and kidney biochemical parameters were used to determine the drug tolerance of mice (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns not significant).

We next asked whether high RRM1 expression in LUAD cells leads to resistance to the antitumor effect of DAC at low doses. To this end, we constructed shRNA to generate RRM1 stable knockdown (KD) in A549 and PC-9 cells, which resulted in about 60% reduction in RRM1 proteins and downregulation of RRM1 RNA levels (Supplementary Fig. S2H, I). Notably, this level of RRM1 knockdown exerted only a modest effect on A549 and PC-9 cell growth (Supplementary Fig. S2J). Conversely, excessive reduction or complete loss of RRM1 expression through knockdown or CRISPR-mediated knockout resulted in cell death, consistent with the essential role of ribonucleotide reductase activity in cell survival and proliferation [25]. Interestingly, RRM1 KD significantly enhanced inhibition of clonogenic survival by DAC (Fig. 2B).

To assess the significance of targeting RRM1 for DAC treatment in vivo, we employed a xenograft model by implanting RRM1 KD and control PC-9 cells to nude mice subcutaneously, followed with DAC treatment at 0.5 mg/kg body-weight three times each week (Fig. 2C). Results showed that either DAC treatment or RRM1 knockdown alone modestly inhibited tumor growth, while the combination resulted in significantly enhanced tumor growth inhibition (Fig. 2D, E), consistent with in vitro results from colony formation assays. Compared with DAC alone or RRM1 knockdown, the combination significantly reduced tumor weight (Fig. 2F). Consistent with the observed reduction in tumor growth, Ki67 immunohistochemistry revealed the lowest proliferation index in RRM1 knockdown tumors treated with DAC compared with all other groups. (Fig. 2G). To evaluate the potential side effects of DAC under this treatment regimen, we found that there was no significant difference in the body weight of mice among the four groups (Fig. 2H), indicating that this dose and frequency of DAC administration were well tolerated by mice. Since the most significant side effect of DAC in clinical use is myelosuppression [26], we collected blood samples from these mice at the end of the drug intervention and examined the peripheral blood cell compositions and biochemical parameters. We found that there were no significant changes in the percentages of major immune cell populations, and biochemical parameters of liver (ALT, AST) and kidney function (creatinine, CREA) (Fig. 2I). These results support that DAC treatment at this low dose, which led to significant tumor growth inhibition in RRM1 knockdown grafts, showed no obvious systemic toxicity. Taken together, RRM1 knockdown sensitizes PC-9 cells to well-tolerated low-dose DAC treatment in xenograft models. These results suggested that RRM1 inhibition sensitizes LUAD cells to DAC treatment in vitro and in vivo.

Pharmacological inhibition of RNR synergizes with DAC treatment to suppress the growth of lung adenocarcinoma cells in vitro and in vivo

Considering RRM1 is the catalytic subunit of RNR, we next investigated whether pharmacological inhibition of RNR could also sensitize LUAD cells to DAC treatment. The RNR inhibitor triapine (3-AP) showed a significant dose-dependent inhibitory effect on the expression of RRM1 protein in A549 and PC9 cells (Supplementary Fig. S3A). While low-dose DAC (50 nM) or 3-AP (200 nM) each showed modest growth inhibition over 4 days, the combination treatment resulted in synergistic growth inhibition—even at this early time point—in three LUAD cell lines (A549, PC-9, and HOP62), as determined by the Excess over Bliss score (Bliss Score = 0 represents additive effect, >0 represents synergy) (Supplementary Fig. S3B). Similarly, the combination treatment of 3-AP and DAC led to synergistic inhibition of colony formation in 3 LUAD cell lines (Fig. 3A). We next evaluated the in vivo anti-tumor effect of combination treatment with DAC and 3-AP (Fig. 3B). In the PC-9 xenograft model, single treatment with DAC or 3-AP exerted an inhibitory effect on tumor growth and weight, but their combination demonstrated a significantly enhanced anti-tumor effect (Fig. 3C–E). Notably, there were no significant differences in the body weight, WBC counts, and blood biochemical tests in the four groups, indicating that combination treatment with DAC and 3-AP is well-tolerated by mice at the doses used (Fig. 3F and supplementary Fig. S3C). Our data suggests that the inclusion of RNR inhibitors may extend the utility of DAC treatment to LUAD.Fig. 3. Pharmacological inhibition of RNR cooperates with DAC treatment to suppress lung adenocarcinoma cell growth in vitro and in vivo.A A549, PC-9, and HOP62 cell lines were seeded into 12-well plates at the rate of 200 cells/ well and treated with 200 nM 3-AP and 50 nM DAC as single agents or in combination. The colony formation ability of A549, PC-9 and HOP62 was shown (n = 3). B and C Female BLAB/c nude mice aged 4–6 weeks were taken in the study. 2 × 10^6^ PC-9-wild type (WT) cells were inoculated subcutaneously. When the tumor volume reached 100–150 mm^3^, mice were randomly divided into 4 groups to start drug administration. The long diameter and short diameter of the tumor were measured every 2–3 days, and the tumor volume was calculated for statistical analysis (n = 5 per group). D and E The long diameter and short diameter of the tumor were measured every 2 days, and the tumor volume and tumor weight were calculated for statistical analysis (n = 5 per group). F The weight of mice was recorded every 2 days and analyzed statistically (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns not significant).

RRM1 knock-down and pharmacological inhibition increase DAC effects on DNMT1 loss and biomarker gene expression

We verified that knockdown and pharmacological inhibition of RRM1/RNR can promote the anti-tumor effect of DAC treatment on LUAD cell lines in vitro and in vivo. DAC causes passive demethylation via incorporation into genomic DNA and covalently trapping DNMT1, thereby promoting its degradation [27]. We explored whether knockdown and pharmacological inhibition of RRM1/RNR can lead to enhanced DAC-mediated loss of DNMT1. We found that DAC treatment in RRM1 KD cells (both A549 and PC-9) decreased DNMT1 protein to a level lower than that of control cells treated with the same dose of DAC (Fig. 4A). Similarly, we examined DNMT1 loss in response to DAC treatment in the presence of 3-AP in A549 and PC-9 cells. Our results showed enhanced DNMT1 loss after combination treatment of DAC and 3-AP (COMB) in comparison to either drug alone (Fig. 4B). Conversely, RRM1 over-expression in the NCI-H1299 cell line, which has low RRM1 background expression, attenuated DNMT1 loss upon DAC treatment (Supplementary Fig. S4A–C). These results suggest that RRM1 mediates resistance to low-dose DAC in lung adenocarcinoma cells by attenuating DAC incorporation and DNMT1 loss.Fig. 4RRM1 knockdown and inhibition increase DAC effects on DNMT1 level and biomarker gene expression*.*A The LUAD cell lines A549 and PC-9 with or without RRM1 knockdown were treated with DMSO or DAC (100 nM) for 6 days. Every 48 h change the medium and drug. Protein samples were extracted, and a western blot was used to detect the expression levels of DNMT1 and RRM1. GAPDH serves as the loading control (n = 3). B The LUAD cell lines A549 and PC-9 were treated with DMSO, DAC (50, 100 nM), and 3-AP (200 nM) for 6 days. Every 48 h change the medium and drug. Protein samples were extracted, and a western blot was used to detect the expression levels of DNMT1 and RRM1. GAPDH serves as the loading control (n = 3). C The LUAD cell lines A549 and PC-9 with or without RRM1 knockdown were treated with DMSO or DAC (100 nM) for 6 days. Every 48 h, change the medium and drug. DNA samples were extracted, and a DNA Dot blot was used to detect the global 5mC level in genomic DNA. Methyl blue serves as the loading control (n = 3). D A549 and PC-9 cell lines with or without RRM1 knockdown were treated with 50 and 100 nM DAC for 6 days. RNA samples were extracted, and qRT-PCR was used to detect the mRNA expression level of the tumor suppressor genes H19 and RASSF1A. GAPDH serves as the internal control (n = 3). E A549 and PC-9 cell lines treated with 200 nM 3-AP and 50 nM DAC were used to extract RNA for qRT-PCR to detect the expression level of H19. GAPDH serves as the internal control (n = 3). F A549 and PC-9 cell lines with or without RRM1 knockdown were treated with 50 and 100 nM DAC for 6 days. Protein samples were extracted, and a western blot was used to detect the expression levels of γ-H2AX. G A549 cell line with or without RRM1 knockdown was treated with 100 nM DAC, and the climbing slices fixed with 4% paraformaldehyde were stained with TdT enzyme and DAPI enzyme, respectively. H Cell cycle analysis of A549 and PC-9 cells treated with DAC for 6 days. Histogram analysis showing the percentage of cell cycle distribution of A549 and PC-9 cells treated with DAC. Three separate experiments were performed for each time point and treatment condition, reported as the mean ± SD (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant).

DAC exerts a well-established passive DNA demethylation effect primarily through inhibiting DNMT1. To investigate whether RRM1 inhibition can further potentiate this DAC-mediated demethylation effect, we performed a DNA dot blot assay to quantify changes in global DNA 5mC levels across different treatment groups. Our results demonstrated that while both the DAC group and RRM1 KD group exhibited a modest reduction in DNA 5mC levels compared to the control, the COMB group showed a significantly more pronounced suppression of 5mC content (Fig. 4C). Notably, this enhanced inhibition of DNA methylation in the COMB group was consistent with the extent of DNMT1 protein downregulation. To explore the epigenetic consequences of attenuated DNMT1 loss, we assessed the mRNA expression level of tumor suppressor genes known to be silenced by abnormal methylation in lung adenocarcinoma [28, 29]. Expression of H19 and RASSF1A was significantly enhanced by DAC treatment with RRM1 knockdown, compared to DAC-treated knockdown control cells (Fig. 4D). Furthermore, the combination treatment of DAC and 3-AP showed significant induction of H19 expression compared to vehicle control or either drug alone in A549 and PC-9 cells (Fig. 4E).

DAC-mediated trapping of DNA methyltransferases (DNMT) to genomic DNA is thought to lead to DNA damage [30]. RRM1 is a key enzyme in nucleoside biosynthesis, DNA replication, and damage repair [31]. It is unclear whether RRM1 inhibition can increase DNA damage alone or in combination with DAC treatment. We found that RRM1 knockdown and DAC treatment alone led to an increase in γH2AX, a DNA damage marker. The combination of RRM1 knockdown and DAC treatment resulted in significantly increased γH2AX (Fig. 4F). DNA damage frequently triggers cellular apoptosis. Consequently, we employed the TUNEL staining assay to investigate whether RRM1 knockdown induces apoptosis and whether combined DAC treatment exacerbates apoptotic responses. TUNEL staining results demonstrated that RRM1 knockdown alone increased apoptotic cell populations, and the combination of RRM1 knockdown with DAC treatment significantly potentiated apoptosis (Fig. 4G). This finding aligns with the intracellular DNA damage profiles, reinforcing the synergistic effect of RRM1 inhibition and DAC in promoting DNA damage-mediated apoptotic pathways. It is reported that DAC can cause G2/M phase arrest [32], and our results showed that the proportion of G2/M phase of A549 and PC-9 cells increased by DAC treatment, either in control or RRM1 knockdown (Fig. 4H). It is also reported that the inhibition of RNR can lead to the arrest of the S phase of cells [33]. Consistently, our results showed a significantly higher proportion of S phase in RRM1knockdown PC-9 cells with DMSO or DAC (Fig. 4H and supplementary Fig. S4D, E). Taken together, RRM1 knockdown may also potentiate the growth inhibitory effect of DAC in A549 and PC-9 cells by influencing the DNA damage and cell cycle arrest.

RRM1 inhibition promotes DAC-induced global epigenetic reprogramming and the immune response pathway

It is reported that high doses of DAC can cause significant cytotoxicity without mediating significant epigenetic effects in the surviving cells, while low doses of DAC can lead to significant DNA demethylation and epigenetic reprogramming of gene expression but without significant cytotoxicity [34–39]. It is unclear whether RRM1 knockdown and DAC treatment combined can promote DAC-induced epigenetic reprogramming of gene expression.

To systematically investigate the transcriptomic response to RRM1 knockdown and DAC treatment in LUAD A549 cells, we collected 4 groups of A549 cells for RNA-seq: control group (Ctrl), DAC treatment group (DAC), RRM1 knockdown group (RRM1 KD), and combination group (Comb, DAC treatment plus RRM1 knockdown). Our results showed that DAC treatment alone has a moderate global epigenetic reprogramming effect on the transcriptome; RRM1 knockdown alone has significantly more differentially expressed genes, while the combination group has the highest number of differentially expressed genes (Fig. 5A and supplementary Fig. S5A–D). To further dissect the underlying molecular mechanisms responsible for the synergistic anti-tumor effect between RRM1 inhibition and DAC, we systematically analyzed the gene ontology (GO) enrichment analysis across five key comparison groups: DAC versus control, RRM1 KD versus control, COMB versus control, COMB versus DAC, and COMB versus RRM1 KD (Supplementary Fig. S6A–D). Notably, to pinpoint the synergism-specific regulatory networks that are exclusively activated by the combined treatment (rather than the individual effects of DAC or RRM1 KD), we constructed a “COMB unique” DEG set. This was achieved by excluding DEGs that were already identified in either the DAC versus control or RRM1 KD versus control groups from the DEGs of the COMB versus control group. Subsequent GO enrichment analysis was specifically performed on this COMB unique gene set, with the results visualized in Fig. 5B.Fig. 5RRM1 inhibition promotes DAC-induced global epigenetic/reprogramming effects and the immune response pathway.A In the RNA-seq experiment, four groups were divided: Control group (Ctrl), DAC treatment group (DAC), RRM1 knock-down group (RRM1-KD), and DAC plus RRM1 knock-down group (COMB). The Volcano plots showed differentially expressed genes between the Comb (unique) group compared to the DAC group. Red: significantly up-regulated genes. Blue: significantly down-regulated genes. B The gene ontology (GO) enrichment analysis across comparison groups: COMB (unique) versus control. C and D A549 with or without RRM1 knockdown were treated with 100 nM DAC for 6 days. Protein and RNA samples were extracted and used to detect protein (C) and mRNA expression levels (D) of interferon-stimulated genes (ISGs). GAPDH serves as the loading control (n = 3). E Heatmaps of significantly dysregulated genes related to the cytosolic DNA-sensing pathway on A549 with or without RRM1 knockdown. F The protein of A549 with or without STING knock out was extracted and used to detect the protein change level of IFI16. GAPDH serves as the loading control (n = 3). KP with or without STING knock out were treated with 200 nM 3-AP for 6 days. Protein and RNA samples were extracted and used to detect protein (H) and mRNA expression levels (G) of interferon-stimulated genes (ISGs). GAPDH serves as the loading control (n = 3). I–L Male C57BL/6 mice aged 4–6 weeks were engrafted subcutaneously with 0.5 × 10^6^ KP cells with or without RRRM1 knockdown. The tumor diameters were measured every 2–3 days, and the tumor volume and tumor weight were calculated for statistical analysis (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant).

As illustrated in Fig. 5B, the COMB unique DEGs exhibited statistically significant enrichment in multiple biological processes and molecular complexes closely associated with the extracellular matrix (ECM). The most prominently enriched terms included “positive regulation of cell adhesion”, “cell-cell adhesion”, and the “NABA CORE MATRISOME”—a curated set of core ECM components. These findings are biologically meaningful, as the ECM serves as a critical structural and signaling scaffold in the tumor microenvironment, and dysregulated cell adhesion is a hallmark of tumor progression, facilitating invasion and metastasis [40, 41]. The enhanced activation of ECM-related gene networks in the COMB group suggests that the synergistic effect of RRM1 inhibition and DAC may, at least in part, be mediated by the modulation of ECM dynamics and cell adhesion properties. Specifically, the upregulation of genes involved in cell adhesion could potentially restrict tumor cell motility and reduce their invasive capacity, thereby contributing to the observed anti-tumor efficacy. We further analyzed which pathways of genes were affected among the four groups (Supplementary Fig. S5E). Compared to other groups, genes related to positive regulation of cell migration were significantly down-regulated in the combination group; while genes related to negative regulation of cell proliferation were significantly up-regulated in the combination group, which is consistent with our previous study that RRM1 inhibition significantly enhanced the anti-tumor proliferation effect of DAC (Supplementary Fig. S5E). Additionally, as reported in previous studies, DAC can significantly upregulate the interferon signaling pathway, thereby regulating the immune response and exerting anti-tumor effects [42, 43]. In our results, compared to the DAC group, immune response-related genes were also significantly increased in the combination group (Supplementary Fig. S5E, F). KEGG pathway analysis also showed enrichment of immune-related signaling pathways, such as the TNF signaling pathway, in the combination group compared to the DAC alone group (Supplementary Fig. S5F). These results showed that RRM1 knockdown significantly enhanced the DAC-induced immune response gene expression programs in A549 cell lines.

Immune-related genes, in addition to tumor suppressor genes, are often silenced in lung adenocarcinomas. Recent studies have reported that stimulator of interferon genes (STING), a crucial adapter in the cytosolic DNA sensing pathway, was silenced by DNA methylation at the promoter region in a considerable portion of lung adenocarcinomas and cell lines, including A549 [44]. It has been suggested that hypermethylation of the STING locus and the resulting inactivation of the cGAS-STING pathway contribute to the poor response to immunotherapy [45]. We thus examined the impact of DAC treatment and RRM1 knockdown on the cGAS-STING pathway activation. DAC treatment alone increased STING expression slightly; whereas the combination of RRM1 KD with DAC treatment substantially elevated STING expression at both RNA and protein levels (Fig. 5C, D). Additionally, this combination activated the type I interferon pathway and interferon-induced genes more substantially than each single-agent treatment. This implicates that lowering RRM1 expression in lung adenocarcinoma may potentiate the immunostimulatory effect of DAC treatment, which is currently under clinical investigation in combination with cancer immunotherapy [46]. Like effects observed with RRM1 knockdown, 3-AP treatment enhanced the ability of DAC treatment in reactivating silenced genes and stimulating STING expression and the related type I IFN response in A549 cells (Supplementary Fig. S5G, H).

Additionally, we found that RRM1 knockdown alone activated the cellular interferon signaling pathway, suggesting the existence of additional signaling mechanisms beyond the dsRNA recognition pathway mediated by DAC-induced demethylation. Re-analysis of RNA-seq data revealed significant upregulation of intracellular DNA recognition signaling pathways following RRM1 knockdown, with notable elevation of the nuclear dsDNA sensor interferon gamma inducible protein 16 (IFI16) protein (Fig. 5E). As reported, the IFI16-STING pathway mediates immune activation during DNA damage [47]. Meanwhile, we also observed that the expression of IFI16 decreased concomitantly with STING knockout (Fig. 5F). This further suggests that IFI16 may play a crucial role in 3-AP-mediated immune activation. To explore whether the combination of RRM1 knockdown and DAC treatment can lead to increased cancer immune response, we turned to a mouse LUAD cell line obtained from immune-competent C57BL/6 background mice, with a KRAS^G12C^ mutation and Trp53 deletion (murine Kras^G12C^/Trp53^−/−^, KP in short) [18]. To validate this, we knocked out STING in KP cells and treated them with an RRM1 inhibitor. STING knockout rescued the immune activation induced by RRM1 inhibition, as evidenced by restored IFN-β1 and CXCL10 mRNA levels (Fig. 5G, H). These results indicate that RRM1 knockdown triggers immune responses via DNA recognition pathways, synergizing with DAC-induced RNA recognition pathway-mediated immunity. We also treated the KP cell line with DAC and 3-AP to find out whether there was a synergistic effect on evoking the immune response in the LUAD cell line. The results showed that the combination group can significantly increase the expression of mRNA levels of type I interferon signaling pathway-related genes in the KP cell line (Supplementary Fig. S5I). The synergistic antitumor effect of the RRM1 inhibitor combined with DAC was further validated in immunocompetent mouse models (Fig. 5I). Results showed that compared with monotherapy, the combination treatment significantly potentiated the inhibition of tumor growth and tumor weight in vivo, and the mice exhibited excellent tolerability to the treatment regimen (Fig. 5J–L and supplementary Fig. S5J). Collectively, RRM1 inhibition promotes DAC-induced global epigenetic reprogramming effects and the immune response pathway.

The differential effects of RNR inhibition on different nucleoside analog DNMT inhibitors

There are two nucleoside analog demethylating agents, azacitidine (AZA) and DAC, that have been approved for MDS or AML by the Food and Drug Administration (FDA) [48, 49]. These two agents are slightly different in structure; AZA is a ribonucleoside while DAC is a deoxyribonucleoside. 80–90% of AZA is integrated into RNA, leading to abnormal ribosome assembly and inhibiting protein synthesis; 10–20% can also be converted into 5-aza-2’-deoxycytidine diphosphate by the action of ribonucleotide reductase and eventually incorporated into genomic DNA, thereby inhibiting DNA methyltransferase and leading to global demethylation. DAC can only be incorporated into genomic DNA, thereby inhibiting DNA methyltransferase and passive demethylation [50, 51]. It is critical to elucidate whether RNR inhibition can have similar effects to AZA as with DAC. In 5 lung cancer cell lines used in DAC incorporation screen, HOP62 cells have the highest mRNA expression of RRM1 (RRM1 mRNA z-score: HOP62, 0.15. A549, -0.19. HCI-H23, -0.23. NCI-H322, -0.56. PC9, -1.44), which is used as a model cell line for treatment with chemical inhibitors of ribonucleotide reductase and potential interaction with AZA or DAC. In lung adenocarcinoma HOP62 cell lines, there is strong synergy in cytotoxicity between DAC and 3-AP in a wide dose range with an overall Bliss score of 1.56 (>0 represents synergism), delayed HOP62 cell growth (Fig. 6A). However, there is strong antagonism in cytotoxicity and growth inhibition between AZA and 3-AP in the same wide dose range with an overall Bliss score of −1.38 (<0 represents antagonism) (Fig. 6B). Consistently, 3-AP and DAC cotreatment significantly delayed HOP62 cells in colony formation compared to either treatment alone, but not in 3-AP and AZA co-treated HOP62 cells (Fig. 6C). We used 3 chemical inhibitors, including 3-AP, hydroxyurea (HU) and clofarabine (CLO), to block the ribonucleotide reductase activity in HOP62 cells and combined with either AZA or DAC. Across the indicated dose range of AZA, all three chemical inhibitors consistently reduced AZA incorporation into genomic DNA (Fig. 6D). On the other hand, all three chemicals consistently increased DAC incorporation into genomic DNA across the indicated dose range (Fig. 6E).Fig. 6RRM1 modulation has differential effects on different nucleoside analog DNMT inhibitors (AZA and DAC).A Inhibition of growth and viability of human lung cancer cell lines HOP62 with DAC and 3-AP. HOP62 cells were treated with the indicated dose range of DAC and 3-AP in the Incucyte time-lapse system. Growth curves were measured as the percent of total confluence in each well over time. The mean percent confluence ± SEM for 16 fields in each condition is shown, with measurements taken every 6 h throughout the course of the growth curve. Viability measurements of cells treated with DAC and 3-AP alone and in combinations across a dose series, as indicated, are represented in a heatmap scaled as shown in the color legend, with each measurement representing the percent viability with respect to the control treatment. The degree of synergy of combinations of DAC and 3-AP was calculated as the Bliss synergy score across the full dose ranges (>0 indicates greater than additive effect). B Inhibition of growth and viability of human lung cancer cell lines HOP62 with azacytidine (AZA) and 3-AP with the same setting as in (A). C HOP62 cells were seeded into 12-well plates, and both DAC + 3AP and AZA + 3AP were used as combined agents to observe the colony formation ability. D, E 5-aza-dC incorporation into genomic DNA, determined by LC–MS/MS in HOP62 cells treated with vehicle control, 800 nM hydroxyurea (HU), 400 nM 3-AP, 100 nM clofarabine (CLO), AZA (D) alone or DAC alone (E) or in combination for 5 days. The mean of the amount of 5-aza-dC per 1000 2-dC in genomic DNA ± SEM of duplicate treatments is shown. Treatment with vehicle control, or RNR inhibitors alone (HU, 3AP, CLO) results in no detection of 5-aza-dC in genomic DNA. F Model of MOA of RNR inhibition and DAC (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant).

Our results are summarized in a proposed model of interaction between RNR and nucleoside analog demethylating agents as shown with arrows (Fig. 6F). RNR inhibitors (3-AP, CLO, HU), or RRM1 knockdown, decrease the reduction and production of dCDP, thus increase the relative proportion and availability of 5-aza-dCDP in the free nucleoside pool after DAC exposure, increase the incorporation of 5-aza-dCTP in genomic DNA and enhance the downstream effects of DAC. On the other hand, RNR inhibitors (3-AP, CLO, HU) directly inhibit the reduction and production of 5-aza-dCDP from 5-aza-CDP after AZA exposure and thus decrease the incorporation of 5-aza-dCTP in genomic DNA and the downstream effects of AZA. Collectively, our data support that RRM1 inhibition may exhibit differential interaction with DAC or AZA.

Discussion

Lung cancer continues to have the highest incidence and mortality; so, it is imperative to develop new therapeutic strategies [1]. DAC is similar in structure to the cytosine nucleoside and thus can be incorporated into genomic DNA during replication. However, DAC does not have therapeutic efficacy in solid tumors [52]. Several clinical studies have found that patients with low RRM1 expression show better progression-free survival (PFS) when treated with gemcitabine, which is also a nucleoside analog drug [53, 54]. We show that RRM1 expression is a key factor limiting the effect of DAC in solid tumors. Previous studies have reported that RRM1 is highly expressed in a variety of cancer tissues, and patients with high expression levels of RRM1 have a poor prognosis [55, 56]. Our data found that the low incorporation rate of DAC was associated with the high expression of RRM1 in lung adenocarcinoma. RRM1 inhibition elevated the incorporation rate of DAC into genomic DNA. We further showed that knocking down RRM1 or inhibiting RNR by 3-AP was sufficient to elevate the DAC incorporation rate through lowering the production and amount of dCDP and dCTP, which indirectly promoted the utilization of DAC for genomic DNA synthesis.

DAC has been suggested to exhibit different anti-tumor properties at high doses and low doses in previous studies [57]. DAC mainly exerts a demethylation effect rather than cytotoxicity at low doses. Significantly, targeting RRM1 or RNR activity potentiated the effect of low doses of DAC treatment in terms of reactivating silenced tumor suppressor genes and showed a strong ability to inhibit tumor growth and proliferation both in vivo and in vitro in our research. In contrast, DNA damage and cell cycle G2/M phase arrest played a major role in anti-tumor effects when high dose levels of DAC were used, indicating cytotoxicity [58, 59]. Consistently, we found that when increasing DAC concentration over a certain threshold, the depletion of DNMT1 in A549 and PC-9 cells became less profound. Interestingly, RRM1 KD not only enhanced the demethylation effect of low-dose DAC, for instance, by reactivating tumor suppressor genes, but also cooperated with low-dose DAC to generate cytotoxicity, including causing obvious DNA damage and G2/M cell cycle arrest. Consequently, the demethylation effect and cytotoxicity caused by DAC and RRM1 inhibition may mediate tumor growth inhibition in the xenograft models in our study.

While TUNEL assay results confirm that RRM1 knockdown synergizes with DAC to enhance apoptosis, the specific apoptotic pathway underlying this effect warrants further discussion. Prior studies showed that RRM1 knockdown disrupts dNTP synthesis, triggering replication stress and DNA double-strand breaks (DSBs) [60]; DAC further amplifies DNA damage, leading to p53 activation. Together, these signals upregulate pro-apoptotic Bcl-2 family proteins and downregulate anti-apoptotic counterparts, promoting mitochondrial outer membrane permeabilization (MOMP) and cytochrome c (Cytc) release [61, 62]. At present, direct evidence for the activation of these pathways is lacking in our study, which represents a limitation to be addressed in future work.

It has been widely reported that DACs reshape the tumor immune microenvironment, thereby improving the immune response to tumors and potentially sensitizing them to immunotherapy, in which the cGAS-STING signaling pathway plays a crucial role [63]. In at least half of melanoma cell lines, DAC has been shown to restore functional STING signaling by inhibiting DNA methylation, enhancing immune recognition, including through upregulating MHC I molecules [64]. Hypermethylation of the STING promoter was reported in KRAS-LKB1-mutated lung cancer cell lines, and treatment with the DNMT inhibitor DAC rescued the expression of STING in these cells [65, 66]. Of note, DAC treatment plays multiple roles, including triggering viral mimicry in the tumor microenvironment and mitigating CD8^+^ T cell exhaustion, which improves tumor responsiveness to anti-PD-1 therapy [46, 67]. In addition to the immune activation effect exerted by DAC, our RNA-seq data analysis revealed that RRM1 inhibition alone may also activate antitumor effects via the IFI16-STING signaling pathway. As a well-characterized nuclear/cytosolic DNA sensor [47], we propose that IFI16 may bind to cytosolic damaged DNA fragments induced by RRM1 inhibition, and subsequently interact with STING to form a functional signaling complex. It is important to clarify that this sequence of molecular events, while consistent with our current data and prior reports, remains a proposed model at this stage and has not been validated by direct experimental evidence. Notably, this observation aligns with prior studies, which have reported that IFI16 expression is often suppressed by aberrant methylation in tumor contexts [68]. Therefore, we speculate that DAC may contribute to the synergistic enhancement of this immune activation pathway through its demethylation activity: specifically, DAC could reverse IFI16 hypermethylation to reactivate IFI16 expression, thereby reinforcing the IFI16-STING signaling cascade initiated by RRM1 inhibition. However, as mentioned above, the demethylation effect of DAC in solid tumors is limited by the incorporation efficiency in genomic DNA. In our data, combining RRM1 knockdown and DAC treatment in A549 cells significantly enhanced immune response pathway expression compared to DAC alone.

In summary, we identified high expression of RRM1 in tumor cells as a limiting factor to the antitumor effect of DAC. Inhibition of RRM1 can promote DAC incorporation into genomic DNA and potentiate the demethylating effect of low-dose DAC, which suppresses lung adenocarcinoma growth through multiple mechanisms. Thus, our findings nominate the combination of RNR inhibitors with DAC, and potentially with further combination with immune checkpoint inhibitors, for clinical translation in lung adenocarcinoma.

While we observed that RRM1 depletion induces genomic instability coupled with DNMT1-dependent RNA pathway dysregulation, the immunological consequences of this dual activation remain unclear. Building on established mechanisms where DNA damage synergizes with RNA-sensing pathways to elicit viral mimicry responses [69], our findings suggest that analogous crosstalk between RRM1-mediated DNA repair defects and endogenous retroviral element activation may underlie the observed immunostimulatory effects. Systematic characterization of cytosolic nucleic acid sensors in this context is warranted. Moreover, although we validated the enhanced antitumor immunity of RRM1 inhibition combined with DAC in immunocompetent mouse models, the lack of combination with immune checkpoint inhibitors represents a limitation of this study. To further advance the clinical translation of this combination therapy, exploring its synergistic effect with immunotherapy in immunocompetent animal models is imperative.

In conclusion, this study identified high RRM1 expression in lung adenocarcinoma cells as a key factor limiting the efficacy of DAC. Inhibition of RRM1, either through genetic knockdown or pharmacological targeting, enhanced DAC incorporation into genomic DNA. This potentiated the demethylating effect of low-dose DAC, leading to the reactivation of tumor suppressor genes, increased DNA damage, and activation of immune response pathways. Mechanistically, RRM1 inhibition reduced dCTP availability, promoting DAC incorporation and DNMT1 depletion while activating the IFI16-STING signaling pathway. Notably, RRM1 inhibition synergized selectively with DAC but not azacitidine, due to differential dNTP pool regulation. These findings suggest that the combination of RNR inhibitors with DAC represents a promising strategy for lung adenocarcinoma treatment, warranting further clinical investigation.

Supplementary information

Supplemental figure legends and tables Supplementary Figure 1 Supplementary Figure 2 Supplementary Figure 3 Supplementary Figure 4 Supplementary Figure 5 Supplementary Figure 6 Full uncropped Gels and Blots image