Radiation induces an EGFR-dependent lactate-epigenetic program underlying synergy with EGFR inhibition

Abstract

Purpose

Clinical studies have suggested a therapeutic synergy between EGFR-TKIs and radiotherapy; however, the underlying mechanisms remain incompletely understood, particularly in tumors harboring the acquired resistance mutation EGFR T790M.

Methods

Using EGFR T790M-positive cell lines, patient-derived organoids (PDOs), and patient-derived xenograft (PDX) models, we systematically investigated radiation responses in EGFR T790M-mutant tumors. Metabolic profiling, protein interaction and phosphorylation analyses, enzyme activity assays, chromatin immunoprecipitation sequencing (ChIP-seq), and DNA damage repair assessments were performed to elucidate the molecular mechanisms of therapeutic synergy between EGFR-TKIs and radiotherapy.

Results

We show that EGFR T790M drives a radiation-amplified glycolytic program characterized by enhanced intracellular lactate accumulation. Radiation induces EGFR-dependent tyrosine phosphorylation and activation of lactate dehydrogenase A (LDHA), sustaining lactate production. Rather than being rapidly exported or oxidized, lactate is retained intracellularly and promotes histone H3 lysine lactylation (H3Kla) at promoter regions of DNA repair genes, establishing a repair-permissive chromatin state. This epigenetic priming enhances repair complex assembly, accelerates recruitment of repair factors to DNA double-strand breaks, and facilitates efficient DNA damage resolution. Genetic or pharmacological disruption of the LDHA-lactate axis impairs DNA repair and restores radiosensitivity. In EGFR T790M PDX models, targeting this metabolic pathway significantly enhances radiotherapy efficacy with acceptable toxicity.

Conclusions

Our findings identify a previously unrecognized metabolic-epigenetic-repair axis through which EGFR T790M promotes radiotherapy resistance. This study redefines the functional role of EGFR T790M in radiation biology, provides mechanistic insight into the clinical synergy between EGFR inhibition and radiotherapy, and offers a rational framework for developing metabolism-informed combination radiotherapy strategies.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s13402-026-01238-x.

Introduction

Non-small cell lung cancer (NSCLC) remains the leading cause of cancer-related morbidity and mortality worldwide [1]. Molecular profiling has revealed that mutations in the epidermal growth factor receptor (EGFR) are of major clinical significance in NSCLC, with particularly high prevalence in Asian populations. EGFR tyrosine kinase inhibitors (EGFR-TKIs) have therefore become the standard of care for patients with EGFR-mutant NSCLC, leading to significant improvements in progression-free survival and overall survival [2]. Among these agents, third-generation EGFR-TKIs such as Osimertinib have substantially improved systemic disease control, particularly in tumors harboring the acquired resistance mutation EGFR T790M [3]. However, despite these advances, inadequate local control, residual lesions, and local or oligoprogressive disease remain frequent clinical challenges in real-world practice [4, 5]. As a result, radiotherapy continues to play a vital role in the management of EGFR-mutant NSCLC, either for primary tumor control or for treatment of lesions that persist or progress during targeted therapy [5, 6]. Nevertheless, despite substantial advances in systemic targeted therapy, inadequate local tumor control and the inevitable development of acquired resistance continue to limit long-term therapeutic efficacy [7]. Radiotherapy represents a cornerstone of local treatment for NSCLC and is widely applied for primary tumor control, management of oligometastatic disease, and treatment of lesions that progress or persist under targeted therapy [8–11]. With the evolution of treatment paradigms, an increasing proportion of patients with EGFR-mutant NSCLC receive radiotherapy in combination with, or sequentially following, EGFR-TKIs at different stages of disease. Evidence from multiple retrospective clinical analyses and exploratory prospective studies indicates that the integration of EGFR-TKIs with radiotherapy can improve local control and, in selected patients, prolong progression-free survival [12–14]. These clinical observations suggest that EGFR signaling may play a critical regulatory role in tumor responses to ionizing radiation.

During EGFR-targeted therapy, the emergence of acquired resistance is almost inevitable, among which the EGFR T790M mutation represents the most common and clinically relevant mechanism [15]. Although third-generation EGFR-TKIs effectively suppress EGFR T790M mutant tumors, patients harboring EGFR T790M frequently require radiotherapy due to local progression or residual disease in real-world clinical practice [4]. Consequently, elucidating the biological characteristics of radiotherapy responses in the context of EGFR T790M is of substantial clinical importance.

Previous studies have demonstrated that EGFR signaling participates in radiation-induced DNA damage responses, including the regulation of DNA repair processes and cell survival pathways [16]. Radiation-induced EGFR activation and subsequent downstream signaling alterations have been closely associated with radiosensitivity [17]. However, most existing studies have focused primarily on signal transduction pathways or individual DNA repair proteins, while the impact of radiotherapy as a systemic stressor on global tumor metabolic states and their downstream consequences has received limited attention. In particular, the intrinsic mechanisms governing radiotherapy responses under the specific molecular background of EGFR T790M remain insufficiently characterized.

Metabolic reprogramming is a fundamental adaptive strategy by which tumor cells survive under hostile microenvironmental conditions and therapeutic stress. EGFR-mutant lung cancers are typically characterized by enhanced glycolytic activity and increased lactate production [18]. In the context of radiotherapy, however, lactate has long been regarded merely as a metabolic byproduct or a contributor to tumor microenvironment acidification, and its potential intracellular signaling or regulatory functions remain poorly understood.

Beyond its conventional role as a metabolic byproduct, lactate is increasingly recognized as a signaling metabolite with the capacity to directly influence epigenetic regulation [19, 20]. The identification of histone lysine lactylation has provided a new conceptual link between intracellular lactate accumulation and chromatin-based gene regulation [21]. However, whether this metabolite-responsive epigenetic modification contributes to radiation-associated DNA damage responses, particularly in EGFR T790M-mutant NSCLC, remains unclear.

Based on these considerations, the present study focuses on EGFR T790M as a central molecular determinant to systematically investigate how this mutation modulates tumor metabolic states and DNA damage responses under radiotherapy. By employing multi-model and multi-layered experimental approaches, we aim to uncover the intrinsic biological basis of radiotherapy tolerance in EGFR T790M-mutant NSCLC and to provide a novel theoretical framework for optimizing radiotherapy-based combination strategies in EGFR-mutant lung cancer.

Method

Cell line origin, authentication, and culture

The human non-small cell lung cancer (NSCLC) cell lines H1975, PC9, and A549 were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). H1975 cells harbor the EGFR L858R and EGFR T790M mutations, PC9 cells carry an EGFR exon 19 deletion, and A549 cells are EGFR wild-type.

The PC9-EGFR T790M stable cell line was established in our laboratory by lentiviral transduction. Briefly, parental PC9 cells were infected with lentiviral particles carrying the EGFR T790M construct (GeneChem, Shanghai, China) at a multiplicity of infection (MOI) of 10, followed by puromycin selection to generate a stable transduced cell population. Successful establishment of the PC9-EGFR T790M cell line was confirmed before subsequent experiments.

H1975 and PC9 cells were cultured in DMEM medium (Gibco,11965092), whereas A549 cells were maintained in RPMI-1640 (Gibco, 11875093); both media were supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. All cells were maintained at 37 °C in a humidified incubator with 5% CO₂.

Cell viability assay and IC50 determination

For 2D cell lines, cells were seeded into 96-well plates at an appropriate density and allowed to attach overnight. The following day, cells were treated with increasing concentrations of Osimertinib or HS-10,296 for 72 h. Cell viability was then measured using the Cell Counting Kit-8 (CCK-8, Yeasen, 40203ES) according to the manufacturer’s instructions. Absorbance was measured using a microplate reader, and viability was normalized to the vehicle-treated control group, which was set as 100%.

For patient-derived organoids (PDOs), organoids were treated with increasing concentrations of Osimertinib or HS-10,296 for the indicated duration. Organoid viability was assessed using the CellTiter-Glo^® 3D Cell Viability Assay (Promega, G9681) according to the manufacturer’s instructions. This assay quantifies intracellular ATP as an indicator of metabolically active viable cells. Luminescence signals were measured and normalized to the vehicle-treated control group.

Dose-response curves were generated using GraphPad Prism version 10 by nonlinear regression with the model “log(inhibitor) vs. normalized response-variable slope (four parameters).” IC50 values were calculated from the fitted curves. For samples that did not reach 50% inhibition within the tested concentration range, the IC50 was considered not reached (NR) and was therefore not estimated.

Irradiation conditions and drug treatment

Ionizing radiation (IR) was delivered using a clinical linear accelerator (LINAC, 6 MV photons) at the Department of Radiation Oncology, Zhongshan Hospital, Fudan University. Cells and organoids were irradiated at room temperature under ambient conditions. Culture plates were positioned at a fixed distance from the radiation source to ensure uniform dose distribution. Sham-irradiated controls were handled identically without radiation exposure.

For clonogenic survival assays, cells were exposed to graded radiation doses of 0, 2, 4, and 6 Gy.

For mechanistic studies, including immunofluorescence, comet assay, and molecular analyses, a single radiation dose of 2 Gy was used unless otherwise indicated.

Patient-derived organoids were irradiated under the same conditions at a dose of 6 Gy, as higher radiation doses are required to achieve measurable responses in 3D organoid systems.

For in vivo irradiation, tumor-bearing mice were treated using the same clinical linear accelerator (LINAC, 6 MV photons). Mice were anesthetized with isoflurane and positioned to allow localized irradiation of the tumor area, while the rest of the body was shielded with lead to minimize off-target exposure. A fractionated radiation regimen of 2 Gy per fraction for 5 consecutive fractions (total dose: 10 Gy) was applied.

Osimertinib (MedChemExpress, HY-15772) and HS-10,296 (MedChemExpress, HY-112823) were used as third-generation EGFR tyrosine kinase inhibitors in this study. For subsequent mechanistic and combination experiments, the working concentrations were selected based on dose-response analyses and preliminary optimization experiments. Osimertinib was used at 20 nM for H1975 cells, 50 nM for PC9-EGFR T790M cells, 50 nM for PDO-EGFR T790M organoids, and 300 nM for A549 cells. HS-10,296 was used at 20 nM for H1975 cells, 50 nM for PC9-EGFR T790M cells, 80 nM for PDO-EGFR T790M organoids, and 300 nM for A549 cells.

Clonogenic assay

For clonogenic survival analysis, cells were seeded into 6-well plates at a density of 500 cells per well and allowed to adhere overnight. Cells were pretreated with the indicated drugs for 24 h prior to irradiation according to the experimental design. After treatment, cells were maintained under continuous drug exposure where applicable and cultured for 10–14 days to allow colony formation. Colonies were then fixed with 4% paraformaldehyde, stained with crystal violet, and colonies containing at least 50 cells were counted manually.

The plating efficiency (PE) was defined as the ratio of the number of colonies formed in the untreated control group to the number of cells seeded (500 cells). The surviving fraction (SF) was calculated as the number of colonies in the treated group divided by the product of 500 and the corresponding PE.

Immunofluorescence

Cells were exposed to ionizing radiation and fixed at 0.5, 2, 6, and 24 h post-irradiation with 4% paraformaldehyde. After fixation, cells were permeabilized with 0.1% Triton X-100 and blocked with 5% bovine serum albumin (BSA) to reduce nonspecific binding. Cells were then incubated overnight at 4 °C with the following primary antibodies: γH2AX (Cell Signaling Technology, #2577, 1:200), PRKDC (Cell Signaling Technology, #12311, 1:200), and 53BP1 (Cell Signaling Technology, #4937, 1:200).

After washing with PBS, cells were incubated with the appropriate fluorophore-conjugated secondary antibodies, including anti-rabbit IgG (Cell Signaling Technology, #4412 or #4413, 1:100) and anti-mouse IgG (Cell Signaling Technology, #4408 or #4409, 1:100), depending on the specific experimental design. Nuclei were counterstained with DAPI. Fluorescence images were acquired using an Olympus confocal microscope equipped with a 63× water-immersion objective, and identical acquisition settings were applied across all experimental groups.

For each condition, representative fields were randomly selected, and foci numbers were quantified in DAPI-positive nuclei using ImageJ/FIJI. At least 50 nuclei per condition were analyzed for foci quantification.

Comet assay

DNA damage was evaluated using an alkaline comet assay with the OxiSelect™ Comet Assay Kit, 3-Well (Cell Biolabs, STA-350) according to the manufacturer’s instructions, with minor modifications. Briefly, at the indicated time points after treatment, cells were harvested, washed once with ice-cold PBS, and resuspended in cold PBS at an appropriate density. Cell suspensions were then mixed with OxiSelect™ Comet Agarose and immediately applied onto precoated OxiSelect™ 3-Well Comet Slides. After gel solidification at 4 °C in the dark for 15–30 min, slides were immersed in prechilled lysis buffer at 4 °C in the dark for 30–60 min. The slides were subsequently transferred to alkaline solution and incubated at 4 °C in the dark for 30 min to allow DNA unwinding.

For alkaline electrophoresis, slides were placed in prechilled alkaline electrophoresis buffer and subjected to electrophoresis at 1 V/cm for 15–30 min, with the buffer volume adjusted to maintain a working current of approximately 300 mA. After electrophoresis, slides were rinsed three times with prechilled deionized water for 2 min each, immersed in 70% ethanol for 5 min, and then dried at 37 °C for 30 min. DNA was stained with Vista Green DNA Dye and comet images were captured using a high-content imaging system with a 10× objective under the red fluorescence channel.

Comet images were analyzed using CometScore software, and at least 50 cells per sample were quantified. Tail moment was used as the primary indicator of DNA damage. This assay was performed as previously described with minor modifications [22, 23].

Glucose uptake and lactate quantification

Cellular glucose uptake was assessed using the fluorescent glucose analog 2-NBDG (Invitrogen, Thermo Fisher Scientific, #N13195). Briefly, the 2-NBDG stock solution was diluted in glucose-free medium to a working concentration of 100 µM and added to the cells. Cells were then incubated at 37 °C for 15–30 min in the dark, followed by extensive washing with PBS to remove excess probe. Intracellular fluorescence was imaged using a high-content imaging system with a 20× objective under the green fluorescence channel, and the fluorescence intensity was used as an indicator of relative glucose uptake.

Lactate levels were measured using the L-Lactate Assay Kit (Colorimetric) (Abcam, #ab65331) according to the manufacturer’s instructions. For intracellular lactate measurement, cells were collected and lysed, and the resulting lysates were subjected to colorimetric detection. For extracellular lactate measurement, culture supernatants were collected and directly analyzed. Absorbance was measured at 450 nm using a microplate reader. To account for differences in cell abundance among groups, both intracellular and extracellular lactate signals were normalized to the corresponding cell number. Lactate levels were not converted into absolute concentrations, but were instead expressed as relative values normalized to the corresponding control group.

Seahorse metabolic flux analysis

Cellular metabolic flux was analyzed using the Agilent Seahorse XFe96 Analyzer. Cells were seeded into Seahorse XF96 cell culture microplates at a density of 5,000 cells per well and cultured for 24 h to allow attachment before the assay. On the day of the experiment, cells were washed and replaced with Seahorse XF assay medium and then incubated in a 37 °C non-CO₂ incubator for 60 min for equilibration prior to measurement.

For the Mito Stress Test, the assay medium consisted of Seahorse XF DMEM medium (pH 7.4) supplemented with 10 mM glucose, 1 mM pyruvate, and 2 mM glutamine. OCR was measured following sequential injection of oligomycin, FCCP, and rotenone/antimycin A. The final concentrations in the wells were 1.5 µM oligomycin, 0.5 µM FCCP, and 0.5 µM rotenone/antimycin A, respectively.

For the Glycolysis Stress Test, ECAR was measured using Seahorse assay medium prepared according to the kit instructions. In brief, the glycolysis stress assay medium did not contain glucose at baseline, and ECAR was recorded following sequential injection of glucose (10 mM), oligomycin (1 µM), and 2-deoxy-D-glucose (2-DG; 50 mM).

Metabolic parameters were calculated using Seahorse Wave software (version 2.6.3). Both ECAR and OCR values were normalized to the corresponding cell number, as indicated. These assays were used to assess glycolytic activity and mitochondrial respiration under different experimental conditions. The experiments were performed according to the manufacturer’s instructions with minor modifications and as previously described [24].

Western blot

Total cellular proteins were extracted using RIPA Lysis Buffer (Epizyme Biotech, #PC102) supplemented with protease and phosphatase inhibitors according to the manufacturer’s instructions. Protein concentrations were determined using a BCA Protein Assay Kit (Epizyme Biotech, #ZJ101). Equal amounts of protein (20 µg per lane) were separated by SDS-PAGE and subsequently transferred onto PVDF membranes. Membranes were blocked with 5% non-fat milk or 5% bovine serum albumin (BSA), depending on the target protein, and then incubated overnight at 4 °C with primary antibodies against H3Kla (Sigma-Aldrich, #SAB5701137, 1:1000), H3 (Cell Signaling Technology, #9715, 1:1000), or other indicated proteins. Membranes were then incubated with the appropriate HRP-conjugated secondary antibodies (Cell Signaling Technology, #7074 or #7076, 1:1000).

Protein bands were visualized using an enhanced chemiluminescence (ECL) detection reagent (Epizyme Biotech, SQ201). Band intensities were quantified using ImageJ software, and the expression levels of target proteins were normalized to the corresponding loading controls.

Proximity ligation assay (PLA)

The proximity ligation assay (PLA) was performed using the Duolink^® PLA kit (Merck, #DUO92101) according to the manufacturer’s instructions to detect close-proximity interactions between EGFR and LDHA or PRKDC and RAD51. Briefly, cells were fixed, permeabilized, and incubated with primary antibodies raised in different species against the indicated target proteins. For detection of the EGFR-LDHA interaction, the primary antibodies used were anti-EGFR (Cell Signaling Technology, #4267S, 1:300) and anti-LDHA (Santa Cruz Biotechnology, #sc-137243, 1:300). For detection of the PRKDC-RAD51 interaction, the primary antibodies used were anti-PRKDC (Cell Signaling Technology, #12311, 1:300) and anti-RAD51 (Santa Cruz Biotechnology, #sc-377467, 1:300). Species-specific PLA probes were then applied, followed by ligation and rolling-circle amplification according to the manufacturer’s protocol.

PLA signals were visualized as discrete red fluorescent puncta, and nuclei were counterstained with DAPI. Images were acquired using a high-content imaging system equipped with a 63× water-immersion objective under the red fluorescence channel. The number of PLA puncta per cell was quantified to assess the relative proximity of the indicated protein pairs under different experimental conditions. This assay was performed as previously described with minor modifications [25].

Phos-tag SDS-PAGE and phosphatase treatment

The phosphorylation status of LDHA was analyzed using Phos-tag SDS–PAGE, which allows separation of proteins according to their phosphorylation-dependent mobility shifts. Briefly, cells were washed with prechilled PBS and lysed on ice using a phosphorylation-preserving lysis buffer supplemented with protease inhibitors. For samples intended for direct Phos-tag analysis, phosphatase inhibitors were additionally included to preserve endogenous phosphorylation signals. Lysates were incubated on ice, clarified by centrifugation at 4 °C, and the supernatants were collected for protein quantification. A total of 50 µg protein per lane was resolved on SuperSep™ Phos-tag™ precast gels (50 µmol/L, 12.5%) (FUJIFILM Wako, #195-17991). Following electrophoresis, proteins were transferred onto PVDF membranes and immunoblotted with an anti-LDHA antibody (Santa Cruz Biotechnology, #sc-137243, 1:300). Membranes were then incubated with the appropriate HRP-conjugated secondary antibody (Cell Signaling Technology, #7076, 1:1000).

Protein bands were visualized using an enhanced chemiluminescence detection system. Distinct mobility-shifted LDHA bands were interpreted as different phosphorylation states. For phosphatase validation, parallel cell lysates were prepared in the absence of phosphatase inhibitors and then treated with λ-phosphatase (New England Biolabs, #P0753S) according to the manufacturer’s instructions prior to electrophoresis. Phosphatase-treated samples exhibited collapse of the slower-migrating bands, confirming that the mobility shift was phosphorylation-dependent. This assay was performed as previously described with minor modifications [26].

LDH enzymatic activity assay

LDH enzymatic activity was measured using a Lactate Dehydrogenase Activity Assay Kit (Sigma-Aldrich, #MAK066) according to the manufacturer’s instructions. Briefly, cells were lysed, and equal amounts of protein from each sample were subjected to enzymatic activity measurement. The assay is based on the catalytic conversion of pyruvate to lactate, coupled to NADH oxidation, which was monitored spectrophotometrically at 450 nm.

To account for differences in cell abundance among groups, the measured LDH activity was normalized to the corresponding cell number. LDH activity was not calculated as an absolute enzymatic rate, but was instead expressed as relative activity, presented as fold change compared with the corresponding control group.

Generation of LDHA-KO and LDHA mutant cell lines

LDHA knockout cell lines were generated using a CRISPR-Cas9 gene-editing system. Single guide RNAs (sgRNAs) targeting the human LDHA gene were designed and synthesized by GeneChem (Shanghai, China) and cloned into a lentiviral CRISPR vector. Cells were transduced with sgRNA-Cas9 lentivirus and subjected to antibiotic selection to establish stable cell populations. Single-cell clones were isolated and expanded.

For rescue experiments, LDHA-knockout cells were transiently transfected with plasmids encoding either wild-type LDHA (LDHA-WT) or a tyrosine-deficient mutant (LDHA-YF) using Lipofectamine 3000 (Invitrogen, USA) according to the manufacturer’s instructions. The LDHA-YF construct was generated by introducing tyrosine-to-phenylalanine (Y→F) substitutions at the indicated site(s), and the corresponding plasmids were synthesized by Sangon Biotech (Shanghai, China).

ChIP-seq and ChIP-PCR

ChIP assays were performed using the Pierce™ Magnetic ChIP Kit (Thermo Scientific, #26157) according to the manufacturer’s instructions with minor modifications. Briefly, cells were crosslinked with 1% formaldehyde for 10 min at room temperature, and the reaction was quenched by adding glycine to a final concentration of 0.125 M. Cells were then lysed and chromatin was sheared by sonication to an average fragment size of 200–600 bp. For each immunoprecipitation reaction, 5 × 10⁶ cells were used. Chromatin was incubated overnight at 4 °C with 6 µg of anti-H3Kla antibody (PTM BIO, #PTM-1427RM) or normal rabbit IgG as a negative control. Antibody–chromatin complexes were captured using Protein A magnetic beads, extensively washed, and eluted, followed by reversal of crosslinking at 65 °C overnight. DNA was purified using the spin columns supplied with the kit.

For ChIP-seq library preparation, purified ChIP DNA was processed using the VAHTS^® Universal Plus DNA Library Prep Kit for Illumina (Vazyme, #ND617) according to the manufacturer’s instructions for ChIP-DNA-compatible library construction. Briefly, ChIP DNA underwent end preparation and 3′-dA tailing, followed by ligation of Illumina-compatible adapters at 20 °C for 15 min. Adapter-ligated DNA was purified using VAHTS DNA Clean Beads (Vazyme, #N411) at 0.6× bead-to-sample volume ratio. Libraries were then amplified using the kit-provided VAHTS HiFi Amplification Mix and PCR Primer Mix 3 for Illumina, with cycle numbers determined according to the recovered DNA amount following the manufacturer’s recommendations. PCR products were purified using VAHTS DNA Clean Beads at 0.9× bead-to-sample volume ratio. Library size distribution was assessed using an Agilent 2100 Bioanalyzer or an equivalent instrument, and effective library concentration was determined using a qPCR-based library quantification method. Libraries were sequenced on an Illumina NovaSeq 6000 platform (paired-end 50 bp) with a target depth of 30–40 million reads per sample.

Sequencing reads were aligned to the human reference genome (hg38) using Bowtie2 (v2.4.1) with default parameters. Duplicate reads were removed using Picard MarkDuplicates. Peaks were called using MACS2 (v2.2.7.1), with input DNA used as background control. Peaks were annotated to the nearest genes using HOMER (v4.11) or ChIPseeker (v1.32). Peaks with q-value < 0.05 and fold enrichment > 2 were considered statistically significant.

ChIP-PCR was performed to validate the enrichment of H3K18la at specific promoter regions of DNA repair genes. Following ChIP, DNA was extracted from the immunoprecipitated chromatin using the Pierce™ Magnetic ChIP Kit (Thermo Scientific, #26157) according to the manufacturer’s instructions. PCR reactions were carried out in a total volume of 25 µL containing 5 µL of eluted DNA, 12.5 µL of 2× Taq PCR Master Mix, and 200 nM of each primer. The PCR amplification was performed according to the manufacturer’s protocol(Yeasen, #MP05501S). PCR products were separated by electrophoresis on a 2% agarose gel containing GelRed in 0.5× TAE buffer at 120 V for 30 min. Gel images were captured using a gel documentation system. Normal rabbit IgG was used as a negative control, and input chromatin (diluted 1:10) served as a positive control. Each ChIP reaction was performed in three independent biological replicates.

RNA-seq and RT-qPCR

Total RNA was extracted from cells using TRIzol reagent (Invitrogen). RNA quality was assessed by NanoDrop and Agilent Bioanalyzer 2100. Libraries were constructed from 1 µg total RNA using the NEBNext^® Ultra™ RNA Library Prep Kit for Illumina^® (NEB, USA) according to the manufacturer’s instructions and sequenced on an Illumina NovaSeq platform to generate 150-bp paired-end reads, with a depth of approximately 30 million reads per sample. Clean reads were aligned to the human reference genome (GRCh38/hg38) using HISAT2 v2.0.5, and gene counts were obtained using featureCounts v1.6.0. Differential expression analysis was performed using DESeq2 with three biological replicates per group, and genes with an adjusted P < 0.05 were considered differentially expressed after Benjamini–Hochberg correction. GSEA was performed using Broad Institute GSEA v4.3.2 to assess pathway-level changes, with a particular focus on metabolic pathways.

Total RNA was isolated from cells and reverse-transcribed into cDNA using the reverse transcription kit (Yeasen, #16948ES60) in accordance with the manufacturer’s protocol. Quantitative PCR was subsequently carried out with gene-specific primers. β-actin was used as the endogenous control for normalization. Relative transcript levels were calculated using the 2^{^−ΔΔCt} method and are presented relative to the control group. The sequences of all primers used in this study are provided in Table 1.

Table 1.

Sequences of all primers used in this study

Gene Symbol	Forward primer(5'->3')	Reverse primer(5'->3')
LDHA	GCCCGATTCCGTTACCTGAT	TAGCCCAGGATGTGTAGCCT
HK2	TGACAATGGCGCTCTCTGTT	ACTCCAGTATTGCAGGTTCCA
GLUT1	CTGTCGTGTCGCTGTTTGTG	AAAGATGGCCACGATGCTCA
LDHB	GCCTTCTCTCTCCTGTGCAA	CCTCTTCTTCCGCAACTGGT
PKM2	CACCTGTACCGTGGCATCTT	GCCTTTTGCCCCACTAAGGT
PRKDC	TCGGCTTTCAACAAAAACTGGT	TGGCACCGTCTTTCCTACAC
RAD51	AGAGACCGAGCCCTAAGGAG	GGCAACAGCCTCCACAGTAT
β-actin	AACACAGTGCTGTCTGGTGG	CAGAGTACTTGCGCTCAGGA

Organoid and PDX models

Patient-derived organoids (PDOs) were established from tumor tissues obtained from NSCLC patients and embedded in Matrigel for three-dimensional culture. The organoids were maintained in a serum-free lung adenocarcinoma organoid culture medium using the Lung Adenocarcinoma Organoid Kit (Bozhen Biotech, K2138-LA) according to the manufacturer’s instructions. These PDO models were used for in vitro studies to assess drug responses and treatment efficacy. Bright-field images of organoids were captured at the indicated time points, and the projected area of individual organoids was measured using ImageJ software. For each condition, 50 organoids were analyzed, and the mean organoid area was used for statistical analysis. The same imaging and analysis settings were applied across all groups. Collection and use of human tumor samples were approved by the Ethics Committee of Zhongshan Hospital, Fudan University (Approval No. B2021-230R).

For in vivo studies, patient-derived xenograft (PDX) models were established by subcutaneous implantation of fresh tumor fragments from the same EGFR T790M-mutant NSCLC patient into female NOD mice (6–8 weeks old). All animal experiments were approved by the Institutional Animal Care and Use Committee of Zhongshan Hospital (Approval No. 2021 − 718). Mice were housed in a specific pathogen-free (SPF) facility with free access to food and water under a 12 h light/12 h dark cycle at controlled temperature and humidity.

When tumors reached approximately 150–200 mm³, mice were randomized into the indicated treatment groups (n = 6 mice per group). Osimertinib (5 mg/kg) and Dichloroacetate (50 mg/kg) were administered by oral gavage once daily for 14 consecutive days according to the experimental design. For in vivo irradiation, tumor-bearing mice were anesthetized with isoflurane and positioned to allow localized irradiation of the tumor area, while the rest of the body was shielded with lead to minimize off-target exposure. Ionizing radiation was delivered using a clinical linear accelerator (LINAC, 6 MV photons) at the Department of Radiation Oncology, Zhongshan Hospital, Fudan University. A fractionated radiation regimen of 2 Gy per fraction for 5 consecutive fractions (total dose: 10 Gy) was applied.

Tumor size was measured regularly using calipers, and tumor volume was calculated using the formula: tumor volume = (length × width²)/2. Systemic toxicity was evaluated by serum biochemical analysis, including ALT, AST, BUN, and CK. Humane endpoints included excessive tumor burden, tumor ulceration or necrosis, marked body weight loss, impaired mobility, lethargy, or any other signs of distress requiring euthanasia according to institutional animal care guidelines. All animal experiments were conducted in accordance with institutional ethical guidelines.

Molecular docking and molecular dynamics simulation

Molecular docking and molecular dynamics (MD) simulations were performed to investigate the binding mode and conformational stability of dichloroacetate (DCA) with LDHA. For molecular docking, the three-dimensional crystal structure of LDHA was obtained from the Protein Data Bank (PDB), and docking was carried out using AutoDock Vina. Before docking, water molecules and irrelevant heteroatoms were removed from the protein structure, and hydrogen atoms were added. The binding pocket was defined based on the reported active-site region of LDHA, and the optimal binding pose was selected according to the docking score and interaction pattern.

For MD simulation, the docked DCA-LDHA complex was subjected to all-atom simulation using GROMACS 2020.6. The protein topology was generated with the AMBER99SB-ILDN force field, whereas the ligand topology was generated using Sobtop 1.0 based on the GAFF force field. The complex was solvated in a truncated octahedral TIP3P water box with a minimum distance of 1.0 nm between the solute and the box boundary. Appropriate numbers of Na⁺ and Cl⁻ ions were added to neutralize the system and balance the ionic environment.

Energy minimization was performed using 2500 steps of the steepest descent algorithm followed by 2500 steps of conjugate gradient minimization. The system was then equilibrated by 100 ps of NVT simulation and 100 ps of NPT simulation at 298.15 K. Temperature was controlled using an appropriate thermostat, and pressure was maintained at 101.325 kPa using an appropriate barostat. Long-range electrostatic interactions were calculated using the particle mesh Ewald (PME) method, and the cutoff for nonbonded interactions was set to 1.0 nm. The collision frequency was set to 2 ps, and the integration time step was 2 fs. Production MD simulation was subsequently performed for 100 ns under periodic boundary conditions, and trajectory coordinates were saved every 10 ps.

Trajectory analysis was performed to evaluate the structural stability and dynamic behavior of the DCA-LDHA complex. The root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg), hydrogen bond number, solvent-accessible surface area (SASA), and interaction energy were analyzed to assess conformational changes and binding stability over time.

Statistical analysis

All experiments were independently repeated at least three times, and n indicates the number of biological replicates. Data are shown as mean ± SD. Statistical analyses were carried out using GraphPad Prism version 10. For comparisons between two groups, an unpaired two-tailed Student’s t-test was used. For comparisons involving more than two groups, one-way ANOVA followed by a suitable post hoc test was applied. Differences were considered statistically significant when P < 0.05.

Result

EGFR T790M inhibition improves radiotherapy response and delays radiation-induced DNA repair in EGFR T790M lung cancer models

To characterize the genotype-specific effects of third-generation EGFR inhibitors, we first examined the drug sensitivity of lung cancer models with distinct EGFR backgrounds. Dose-response analyses revealed that H1975 cells harboring endogenous EGFR T790M mutation, PC9-EGFR T790M cells with acquired EGFR T790M mutation, and patient-derived EGFR T790M organoids (PDO-EGFR T790M) exhibited pronounced dose-dependent growth inhibition upon Osimertinib or HS-10,296 treatment, whereas EGFR-wild-type A549 cells were largely insensitive. These results indicate that EGFR T790M mutant tumors are highly dependent on EGFR signaling and represent a sensitive context for Osimertinib/HS-10,296 treatment (Fig. 1a-b). Clonogenic survival assays further demonstrated that Osimertinib or HS-10,296 pre-treatment markedly enhanced the long-term inhibitory effect of radiotherapy in EGFR T790M mutant cells. Combined treatment significantly reduced clonogenic survival in H1975 and PC9-EGFR T790M cells compared with radiotherapy alone, whereas no additional radiosensitization was observed in EGFR-wild-type PC9 or A549 cells, highlighting the genetic specificity of this effect (Fig. 1c-d). To further validate radiosensitization across different radiation doses, clonogenic survival assays were performed using graded irradiation (0, 2, 4, and 6 Gy). Osimertinib treatment resulted in a clear dose-dependent reduction in clonogenic survival in H1975 cells, as reflected by a pronounced downward shift of the survival curve, whereas no significant effect was observed in A549 cells (Supplementary Fig. 1a). These findings confirm a dose-dependent and genotype-specific radiosensitizing effect of Osimertinib in EGFR T790M-mutant cells. To investigate whether radiosensitization was associated with altered DNA damage repair, we analyzed the dynamics of γH2AX foci. Although initial radiation-induced DNA damage was comparable among RT-treated groups, Osimertinib/HS-10,296 combined with radiotherapy markedly delayed γH2AX foci resolution, indicating impaired double-strand break (DSB) repair (Fig. 1e-f). As a genotype-matched control, we further examined DNA damage foci kinetics in EGFR-wild-type A549 cells. Although irradiation induced strong foci formation at early time points, the addition of Osimertinib did not result in a substantial delay in foci clearance compared with radiation alone, indicating that the impaired DNA repair effect of combined treatment is much less evident in EGFR-wild-type cells (Supplementary Fig. 1b). Consistently, analysis of 53BP1 foci revealed sustained persistence of DNA damage sites following combined treatment, suggesting inhibition of non-homologous end-joining (NHEJ) repair, in agreement with the γH2AX findings (Fig. 1g-h). At the functional level, comet assays showed that Osimertinib/HS-10,296 combined with radiotherapy significantly increased the accumulation of unrepaired DNA damage, as reflected by elevated tail moment values (Fig. 1i-j). Finally, in clinically relevant patient-derived organoid models, Osimertinib/HS-10,296 markedly enhanced radiotherapy-induced growth suppression in PDO-EGFR T790M organoids, whereas no significant combinatorial effect was observed in PDO-WT models, confirming the translational potential of this combined treatment strategy (Fig. 1k-l).

Fig. 1.

EGFR T790M inhibition improves radiotherapy response and delays radiation-induced DNA repair in EGFR T790M lung cancer models. (a-b) Dose-response curves of Osimertinib and HS-10,296 in 2D cell cultures and 3D patient-derived organoids. H1975, PC9-EGFR T790M, A549 cells, and patient-derived EGFR T790M organoids (PDO-EGFR T790M) were treated with increasing concentrations of Osimertinib or HS-10,296 for 72 h. Viability of 2D cell lines was measured using the CCK-8 assay, whereas PDO viability was assessed using an ATP-based luminescence assay. Dose-response curves were fitted by nonlinear regression, and IC50 values were calculated accordingly. For A549 cells, no reliable IC50 was reached within the tested concentration range, and the value is therefore presented as NR. (c-d) Clonogenic survival assays evaluating the radiosensitizing effect of Osimertinib or HS-10,296. H1975, PC9-EGFR T790M, PC9, and A549 cells were treated with Osimertinib or HS-10,296 alone, radiation alone, or drug pre-treatment for 24 h followed by radiation, and clonogenic survival was quantified. (e-f) Representative images and quantification of γH2AX foci dynamics following radiation. H1975 and PC9-EGFR T790M cells were subjected to radiation with or without Osimertinib/HS-10,296 pre-treatment, and γH2AX foci were analyzed at indicated time points to assess DNA double-strand break repair kinetics.Quantification was performed on at least 50 cells per condition. (g-h) Analysis of 53BP1 foci dynamics as a readout of non-homologous end-joining (NHEJ) repair following radiation with or without Osimertinib/HS-10,296 pre-treatment in H1975 and PC9-EGFR T790M cells.Quantification was performed on at least 50 cells per condition. (i-j) Comet assays assessing the accumulation of unrepaired DNA damage in H1975 and PC9-EGFR T790M cells treated with radiation alone or Osimertinib/HS-10,296 followed by radiation. DNA damage was quantified by tail moment.Quantification was performed on at least 50 cells per condition. (k-l) Growth inhibition of patient-derived lung cancer organoids following combined treatment. PDO-EGFR T790M and PDO-WT organoids were treated with Osimertinib or HS-10,296 alone, radiation alone, or drug pre-treatment followed by radiation, and organoid area were measured over time. For each condition, 50 organoids were analyzed for projected area quantification. Unless otherwise indicated, data are presented as mean ± SD from at least three independent biological replicates, with n = 3 or n = 6 depending on the specific assay. Technical duplicate or triplicate measurements were averaged within each biological replicate before statistical analysis. For immunofluorescence, PLA, comet assays, and organoid analyses, at least 50 cells or 50 organoids were quantified per condition in each independent biological replicate

EGFR T790M promotes radiotherapy resistance by driving glycolytic reprogramming and radiation-amplified lactate production

To define features associated with EGFR T790M mutation, RNA-seq analysis was performed in EGFR wild-type PC9 and PC9-EGFR T790M cells under basal conditions. Gene set enrichment analysis revealed significant enrichment of glycolysis-related gene signatures in PC9-EGFR T790M cells compared with wild-type cells, indicating an intrinsic upregulation of glycolytic transcriptional programs in the presence of the EGFR T790M mutation (Fig. 2a). At the functional level, glucose uptake assays showed higher basal glucose uptake in EGFR T790M mutant cells. Radiation exposure increased glucose uptake across all tested cell lines, with a more pronounced increase observed in H1975 and PC9-EGFR T790M cells, suggesting elevated metabolic demand under radiation stress (Fig. 2b). Consistently, intracellular lactate measurements demonstrated increased lactate accumulation under both basal and irradiated conditions, with a more pronounced increase observed in H1975 and PC9-EGFR T790M cells (Fig. 2c). Enhanced glycolytic activity was further confirmed by Seahorse analysis. ECAR measurements revealed higher basal glycolysis in EGFR T790M mutant cells and a robust increase following radiation, reflecting enhanced glycolytic capacity (Fig. 2d). At the molecular level, qPCR analysis showed upregulation of key glycolytic enzymes, including GLUT1, HK2, PFKFB3, LDHA, PKM2, and LDHB, supporting the metabolic reprogramming associated with the EGFR T790M mutation (Fig. 2e).To assess the functional relevance of glycolytic enhancement, metabolic inhibition was applied. Pharmacological blockade of glycolysis with 2-deoxyglucose or inhibition of lactate production using an LDHA inhibitor markedly attenuated radiation-induced increases in ECAR and lactate levels, indicating that radiation-associated metabolic changes require an intact glycolysis-lactate pathway (Fig. 2f-g). At the level of DNA damage repair, metabolic inhibition significantly altered post-radiation repair kinetics. γH2AX and 53BP1 foci analyses showed delayed clearance of DNA damage following combined metabolic inhibition and radiation compared with radiation alone. These findings were further supported by comet assays demonstrating increased accumulation of unrepaired DNA damage (Fig. 2h-j).Finally, in patient-derived PDO-EGFR T790M organoid models, inhibition of glycolysis or lactate production markedly enhanced radiation-induced growth suppression. Compared with radiation alone, combined metabolic inhibition resulted in a pronounced reduction in organoid size and structural integrity, confirming the clinical relevance of metabolic reprogramming in radiation response of EGFR T790M mutant tumors (Fig. 2k). To further determine whether blockade of the glycolysis-lactate axis functionally affects radiosensitivity in different EGFR backgrounds, we performed clonogenic survival assays using graded radiation doses in the presence or absence of LDHA inhibition. In EGFR T790M-mutant H1975 cells, LDHA inhibition resulted in a pronounced downward shift of the radiation survival curve, whereas in EGFR-wild-type A549 cells the effect was minimal, indicating that the radiosensitizing effect of lactate-axis inhibition is largely genotype-dependent (Fig. 2l-m).

Fig. 2.

EGFR T790M promotes radiotherapy resistance by driving glycolytic reprogramming and radiation-amplified lactate production. (a) RNA sequencing and gene set enrichment analysis (GSEA) comparing EGFR wild-type PC9 and PC9-EGFR T790M cells under basal conditions. Enrichment of glycolysis-related gene signatures is shown. (b) Glucose uptake measured by 2-NBDG incorporation in H1975, PC9-EGFR T790M, and PC9 cells treated with control conditions or radiation. (c) Intracellular lactate levels in H1975, PC9-EGFR T790M, and PC9 cells under control conditions or following radiation exposure. (d) Extracellular acidification rate (ECAR) assessed by Seahorse glycolysis stress test in H1975, PC9-EGFR T790M, and PC9 cells with or without radiation treatment. (e) Quantitative PCR analysis of glycolysis-related genes, including GLUT1, HK2, PFKFB3, LDHA, PKM2, and LDHB, in H1975, PC9-EGFR T790M, and PC9 cells under control conditions or following radiation. (f-g) Effects of metabolic inhibition on radiation-associated metabolic changes. ECAR (f) and intracellular lactate levels (g) were measured in H1975 and PC9-EGFR T790M cells treated with radiation alone or in combination with 2-deoxyglucose (2-DG, 5 mM, MedChemExpress) or an LDHA inhibitor (FX11, 10 µM, MedChemExpress). (h-i) Analysis of DNA damage repair kinetics following radiation with or without metabolic inhibition. γH2AX (h) and 53BP1 (i) foci were quantified in EGFR T790M LUAD cells treated with radiation alone or combined with 2-deoxyglucose (2-DG, 5 mM, MedChemExpress) or an LDHA inhibitor (FX11, 10 µM, MedChemExpress). Quantification was performed on at least 50 cells per condition. (j) Comet assays assessing residual DNA damage in H1975 and PC9-EGFR T790M cells following radiation with or without 2-deoxyglucose (2-DG, 5 mM, MedChemExpress) or an LDHA inhibitor (FX11, 10 µM, MedChemExpress).Quantification was performed on at least 50 cells per condition. (k) Bright-field images and growth analysis of patient-derived PDO-EGFR T790M organoids treated with control conditions, radiation alone, or radiation combined with 2-deoxyglucose (2-DG, 5 mM, MedChemExpress) or an LDHA inhibitor (FX11, 10 µM, MedChemExpress). For each condition, 50 organoids were analyzed for projected area quantification. (l) Clonogenic survival assay of H1975 cells treated with radiation alone or in combination with an LDHA inhibitor (FX11, 10 µM, MedChemExpress) across graded radiation doses (0, 2, 4, and 6 Gy). (m) Clonogenic survival assay of A549 cells treated with radiation alone or in combination with LDHA inhibition (FX11, 10 µM, MedChemExpress) across graded radiation doses (0, 2, 4, and 6 Gy).Unless otherwise indicated, data are presented as mean ± SD from at least three independent biological replicates, with n = 3 or n = 6 depending on the specific assay. Technical duplicate or triplicate measurements were averaged within each biological replicate before statistical analysis. For immunofluorescence, PLA, comet assays, and organoid analyses, at least 50 cells or 50 organoids were quantified per condition in each independent biological replicate

Radiation enhances EGFR T790M-dependent tyrosine phosphorylation of LDHA to drive glycolytic reprogramming

To identify key metabolic regulators associated with EGFR T790M, RNA-seq data comparing EGFR wild-type PC9 and PC9-EGFR T790M cells were analyzed. Multiple glycolytic enzymes were upregulated in PC9-EGFR T790M cells, among which LDHA exhibited relatively high expression, suggesting its potential involvement in EGFR T790M-associated metabolic reprogramming (Fig. 3a). We next examined the spatial relationship between EGFR and LDHA. Proximity ligation assays revealed minimal EGFR-LDHA signals in EGFR wild-type PC9 cells, whereas clear PLA signals were detected in EGFR T790M mutant H1975 and PC9-EGFR T790M cells. Radiation exposure further increased EGFR-LDHA proximity signals, while Osimertinib treatment alone or in combination with radiation markedly reduced these interactions. Notably, these PLA signals were predominantly localized in the cytoplasm, consistent with the known subcellular distribution of EGFR and LDHA and supporting a cytosolic interaction linked to metabolic regulation (Fig. 3b). To determine the molecular basis of LDHA mobility shifts, immunoblot analysis was performed under different treatment conditions. In EGFR T790M mutant cells, particularly following radiation, LDHA exhibited a mobility shift on Phos-tag SDS-PAGE. This shift was largely abolished by phosphatase or Osimertinib treatment (Fig. 3c). Analysis of public databases revealed multiple putative tyrosine phosphorylation sites on LDHA, providing a structural basis for post-translational regulation, especially at Y10 (Fig. 3d). Consistent with this, in EGFR T790M mutant cells, Phos-tag SDS-PAGE showed multiple species of LDHA with slower migration speed and higher phosphorylation after radiation. Osimertinib significantly inhibited this phenomenon. In contrast, the tyrosine-deficient LDHA mutant (LDHA-yf) also inhibited this phenomenon (Fig. 3e). Functionally, LDH enzymatic activity was elevated in EGFR T790M mutant cells and further increased following radiation, whereas Osimertinib treatment alone or combined with radiation reduced LDH activity (Fig. 3f). In LDHA knockout cells reconstituted with either wild-type or tyrosine-deficient LDHA, restoration of LDHA-WT recovered LDHA activity, glycolytic flux (ECAR), and lactate production, while the LDHA-YF mutant failed to fully restore these metabolic functions (Fig. 3g-i). Analysis of DNA damage repair kinetics revealed that, following radiation, cells rescued with LDHA-WT exhibited progressive clearance of γH2AX foci, whereas cells expressing LDHA-YF displayed prolonged persistence of DNA damage signals. Comet assays further confirmed increased accumulation of unrepaired DNA damage in LDHA-YF-rescued cells (Fig. 3j-k).Finally, in patient-derived PDO EGFR T790M organoid models, inhibition of LDHA in combination with radiation markedly suppressed organoid growth and disrupted organoid structure compared with radiation alone (Fig. 3l).

Fig. 3.

Radiation enhances EGFR T790M-dependent tyrosine phosphorylation of LDHA to drive glycolytic reprogramming. (a) Heatmap of glycolysis-related enzyme expression derived from RNA-seq analysis comparing EGFR wild-type PC9 and PC9-EGFR T790M cells under basal conditions. Representative enzymes from the glycolysis pathway are shown. (b) Proximity ligation assay (PLA) detecting spatial proximity (< 40 nm) between EGFR and LDHA in PC9 (EGFR-WT), H1975, and PC9-EGFR T790M cells treated with control conditions, radiation, Osimertinib, or Osimertinib combined with radiation. Representative images and quantification of PLA signals per cell are shown.PLA signals are predominantly observed in the cytoplasm, consistent with the subcellular localization of EGFR and LDHA.Quantification was performed on at least 50 cells per condition. (c) Phos-tag SDS-PAGE analysis of LDHA electrophoretic mobility in H1975, and PC9-EGFR T790M cells treated with control conditions, radiation, Osimertinib, or Osimertinib combined with radiation, with or without phosphatase treatment, to assess phosphorylation-dependent mobility shifts. (d) Schematic representation of putative tyrosine phosphorylation sites on LDHA based on analysis from the PhosphoSitePlus database. (e) Phos-tag SDS-PAGE analysis of LDHA phosphorylation status in H1975 and PC9-EGFR T790M cells treated with control conditions, radiation, Osimertinib, or Osimertinib combined with radiation. Cells expressing wild-type LDHA or a tyrosine-deficient LDHA mutant (LDHA-YF) were analyzed. (f) LDHA enzymatic activity measured in PC9-EGFR T790M and H1975 cells under control conditions, radiation, Osimertinib treatment, or Osimertinib combined with radiation. (g-i) Functional analysis of LDHA tyrosine phosphorylation using knockout and rescue approaches. LDHA-knockout H1975 and PC9-EGFR T790M cells were reconstituted with vector control, wild-type LDHA, or LDHA-YF mutant, followed by assessment of LDHA enzymatic activity (g), extracellular acidification rate (ECAR) (h), and intracellular lactate levels (i), with or without radiation. (j-k) Analysis of DNA damage repair kinetics in LDHA-knockout cells rescued with wild-type or LDHA-YF mutant following radiation. γH2AX foci dynamics are shown in (j), and comet assays assessing residual DNA damage are shown in (k).Quantification was performed on at least 50 cells per condition. (l) Growth analysis of patient-derived PDO-EGFR T790M organoids treated with radiation alone or radiation combined with an LDHA inhibitor. Representative bright-field images and quantification of organoid size are shown. For each condition, 50 organoids were analyzed for projected area quantification. Unless otherwise indicated, data are presented as mean ± SD from at least three independent biological replicates, with n = 3 or n = 6 depending on the specific assay. Technical duplicate or triplicate measurements were averaged within each biological replicate before statistical analysis. For immunofluorescence, PLA, comet assays, and organoid analyses, at least 50 cells or 50 organoids were quantified per condition in each independent biological replicate

Radiation induces intracellular lactate accumulation that is diverted to histone lactylation in EGFR T790M lung cancer cells

To determine the fate of radiation-induced lactate accumulation, we first performed time-course analyses of intracellular and extracellular lactate levels following radiation exposure. In both H1975 and PC9-EGFR T790M cells, radiation rapidly increased intracellular lactate levels, whereas extracellular lactate exhibited relatively modest changes within the same time window, indicating predominant intracellular lactate accumulation rather than rapid secretion (Fig. 4a). To assess whether extracellular lactate recycling contributes to intracellular lactate pools, culture medium was frequently replaced after radiation to disrupt lactate reuptake. Despite continuous medium exchange, radiation still induced a marked increase in intracellular lactate levels, comparable to that observed under non-medium-change conditions, with only minor attenuation. These results suggest that intracellular lactate accumulation primarily reflects intracellular production and retention rather than extracellular export followed by reuptake (Fig. 4b). We next examined the role of monocarboxylate transporters (MCTs) in lactate transmembrane flux. Pharmacological inhibition of MCTs altered extracellular lactate dynamics and modulated lactate transport under radiation conditions but did not fully prevent radiation-induced intracellular lactate accumulation, indicating that lactate is predominantly generated endogenously while MCT-mediated transport provides an entry point for extracellular lactate manipulation (Fig. 4c). To determine whether accumulated lactate is oxidized via mitochondrial metabolism, oxygen consumption rate (OCR) was measured following radiation. OCR did not increase after radiation exposure and instead showed a decreasing trend, suggesting that intracellular lactate is not substantially cleared through mitochondrial oxidation and is likely diverted to non-oxidative pathways (Fig. 4d). We then examined whether radiation-induced lactate accumulation is coupled to histone lactylation. Western blot analysis revealed a significant increase in histone H3 lysine lactylation (H3Kla) following radiation in both H1975 and PC9-EGFR T790M cells, with a time-dependent pattern, indicating that lactate accumulation is translated into epigenetic modification (Fig. 4e). Finally, to establish a causal relationship between lactate and H3Kla, cells were treated with exogenous lactate. Exogenous lactate markedly increased H3Kla levels, whereas inhibition of MCTs substantially attenuated lactate-induced H3Kla elevation (Fig. 4f).

Fig. 4.

Radiation induces intracellular lactate accumulation that is diverted to histone lactylation in EGFR T790M lung cancer cells. (a) Time-course analysis of intracellular and extracellular lactate concentrations in H1975 and PC9-EGFR T790M cells following radiation exposure (2 Gy). Lactate concentrations were measured at the indicated time points. (b) Assessment of the contribution of extracellular lactate recycling to intracellular lactate accumulation. H1975 and PC9-EGFR T790M cells were subjected to radiation, followed by either standard culture conditions or frequent medium replacement to disrupt lactate reuptake, and intracellular lactate concentrations were measured at the indicated time points. (c) Evaluation of monocarboxylate transporter (MCT)-mediated lactate transport following radiation. Intracellular and extracellular lactate concentrations were measured in H1975 and PC9-EGFR T790M cells treated with control conditions, radiation (2 Gy) alone, or radiation (2 Gy) in combination with an MCT inhibitor (AZD3965, 1 µM, MedChemExpress). (d) Oxygen consumption rate (OCR) measured by Seahorse analysis in H1975 and PC9-EGFR T790M cells under control conditions or following radiation exposure (2 Gy), to assess mitochondrial oxidative metabolism. (e) Western blot analysis of histone H3 lysine lactylation (H3Kla) in H1975 and PC9-EGFR T790M cells treated with control conditions or radiation (2 Gy) at the indicated time points. (f) Western blot analysis of H3Kla levels in H1975 and PC9-EGFR T790M cells treated with exogenous lactate (sodium L-lactate, 10 mM, Sigma-Aldrich) in the presence or absence of an MCT inhibitor (AZD3965, 1 µM, MedChemExpress). Unless otherwise indicated, data are presented as mean ± SD from at least three independent biological replicates, with n = 3 or n = 6 depending on the specific assay. Technical duplicate or triplicate measurements were averaged within each biological replicate before statistical analysis. For immunofluorescence, PLA, comet assays, and organoid analyses, at least 50 cells or 50 organoids were quantified per condition in each independent biological replicate

Lactate-driven histone lactylation epigenetically facilitates PRKDC- and RAD51-mediated DNA repair following radiation

To determine whether radiation reshapes histone lactylation at the genome-wide level, H3Kla ChIP-seq was performed in H1975 cells. Compared with control conditions, radiation markedly increased the total number of H3Kla peaks and altered their distribution across genomic regions, indicating a global redistribution of H3Kla following radiation exposure (Fig. 5a). At the gene level, IGV visualization of ChIP-seq data revealed enhanced H3Kla signals at the promoter regions of DNA repair genes, including PRKDC and RAD51, following radiation, whereas these signals were weak under control conditions (Fig. 5b). These findings were further validated by H3Kla ChIP-qPCR in EGFR T790M models, where radiation consistently increased H3Kla enrichment at the PRKDC and RAD51 promoters (Fig. 5c). We next examined the role of lactate production in this epigenetic modification. Inhibition of LDHA markedly attenuated radiation-induced H3Kla enrichment at DNA repair gene promoters, indicating that lactate generation is required for this chromatin modification (Fig. 5d). At the transcriptional level, RT-qPCR analysis showed only modest increases in PRKDC and RAD51 mRNA expression following radiation, suggesting that H3Kla enrichment is not accompanied by robust transcriptional activation (Fig. 5e). Functionally, proximity ligation assays demonstrated increased PRKDC-γH2AX interactions following radiation, which were further enhanced by exogenous lactate and reduced by LDHA inhibition (Fig. 5f). Consistently, immunofluorescence analysis revealed that lactate treatment promoted recruitment of PRKDC to γH2AX-marked DNA double-strand breaks, as evidenced by both an increase in PRKDC foci per cell and an elevated proportion of γH2AX foci co-localized with PRKDC. In contrast, inhibition of lactate production markedly impaired PRKDC recruitment to DNA damage sites (Fig. 5g). Notably, Osimertinib treatment similarly attenuated PRKDC foci formation following irradiation. Quantitative analysis revealed that while radiation induced a rapid increase in PRKDC foci, peaking at early time points (0.5–2 h), the addition of Osimertinib significantly reduced PRKDC foci across all time points compared with radiation alone (Fig. 5h). Analysis of DNA damage repair kinetics further demonstrated that exogenous lactate accelerated the resolution of γH2AX foci following irradiation, whereas LDHA inhibition resulted in sustained persistence of γH2AX foci at later time points, indicating delayed DNA repair (Fig. 5i). Finally, in patient-derived PDO-EGFR T790M organoids, combined LDHA inhibition and radiation markedly enhanced growth suppression compared with radiation alone, demonstrating the functional relevance of this metabolic-epigenetic regulation axis (Fig. 5j).

Fig. 5.

Lactate-driven histone lactylation promotes recruitment of DNA repair machinery after radiation. (a) Genome-wide profiling of histone H3 lysine lactylation (H3Kla) by ChIP-seq in H1975 cells under control conditions or following radiation (2 Gy, 6 h). Shown are changes in total H3Kla peak numbers and the genomic distribution of peaks across promoter, enhancer, and intergenic regions. (b) Representative IGV tracks showing H3Kla ChIP-seq signals at the promoter regions of DNA repair genes (PRKDC and RAD51) in H1975 cells under control conditions or following radiation. (c) Validation of radiation-induced H3Kla enrichment at DNA repair gene promoters by ChIP-PCR. H3Kla occupancy at the PRKDC and RAD51 promoter regions was quantified in EGFR T790M cells under control conditions or following radiation at the indicated time points. (d) Effects of lactate production inhibition on H3Kla enrichment at DNA repair gene promoters. H3Kla ChIP-PCR was performed in EGFR T790M cells treated with radiation alone or radiation in combination with an LDHA inhibitor. (e) RT-qPCR analysis of PRKDC and RAD51 mRNA expression in H1975 and PC9-EGFR T790M cells under control conditions or following radiation at the indicated time points. (f) Proximity ligation assay (PLA) assessing the assembly of DNA repair complexes. Interactions between PRKDC and RAD51 were quantified in H1975 and PC9-EGFR T790M cells treated with radiation alone, radiation plus exogenous lactate, or radiation plus an LDHA inhibitor.Quantification was performed on at least 50 cells per condition. (g) Immunofluorescence analysis of repair protein recruitment to DNA damage sites. Co-localization of PRKDC with γH2AX-marked DNA double-strand breaks was examined in H1975 and PC9-EGFR T790M cells under the indicated treatment conditions. Quantification was performed on at least 50 cells per condition. (h) Immunofluorescence analysis of of PRKDC foci formation over time following irradiation under the indicated treatment conditions. H1975 were treated with vehicle control, Osimertinib, radiation, or radiation combined with Osimertinib, and PRKDC foci per cell were quantified at 0.5, 2, 6, and 24 h post-irradiation. Quantification was performed on at least 50 cells per condition. (i) DNA damage repair kinetics assessed by γH2AX dynamics in H1975 and PC9-EGFR T790M cells treated with radiation alone, radiation plus lactate, or radiation plus an LDHA inhibitor.Quantification was performed on at least 50 cells per condition. (j) Three-dimensional growth assays evaluating radiation response. Tumor spheroids or patient-derived PDO-EGFR T790M organoids were treated with radiation alone or radiation combined with an LDHA inhibitor, and changes in spheroid or organoid size were quantified. For each condition, 50 organoids were analyzed for projected area quantification. Unless otherwise indicated, data are presented as mean ± SD from at least three independent biological replicates, with n = 3 or n = 6 depending on the specific assay. Technical duplicate or triplicate measurements were averaged within each biological replicate before statistical analysis. For immunofluorescence, PLA, comet assays, and organoid analyses, at least 50 cells or 50 organoids were quantified per condition in each independent biological replicate

Triple targeting of EGFR T790M and lactate metabolism synergizes with radiotherapy: a mechanism-based proof-of-concept strategy

To pharmacologically target the lactate metabolic axis, a mechanism-guided drug repurposing screen was performed using a panel of FDA-approved compounds. Candidate drugs were evaluated in combination with radiation and Osimertinib. Among the six leading candidates tested, Dichloroacetate uniquely exhibited a consistent combinatorial inhibitory effect, whereas the remaining compounds failed to show comparable synergy (Fig. 6a). To assess whether Dichloroacetate directly interacts with LDHA, molecular docking and molecular dynamics simulations were conducted. These analyses revealed that Dichloroacetate stably occupies the LDHA active pocket and maintains a stable binding conformation throughout the simulation period, supporting a direct interaction between Dichloroacetate and LDHA (Fig. 6b). At the metabolic level, we examined the effects of Dichloroacetate on radiation-induced lactate metabolism. In H1975 and PC9-EGFR T790M cells, Dichloroacetate combined with radiation significantly reduced intracellular lactate accumulation induced by radiation (Fig. 6c). Consistently, Seahorse analysis demonstrated that Dichloroacetate markedly attenuated the radiation-induced increase in glycolytic flux, as reflected by reduced ECAR values (Fig. 6d). We next assessed the impact of Dichloroacetate on DNA damage repair. Compared with radiation alone, combined Dichloroacetate and radiation treatment resulted in prolonged persistence of γH2AX signals, indicating delayed DNA damage resolution (Fig. 6e). Immunofluorescence co-localization analysis further showed reduced recruitment of RAD51 to γH2AX-marked DNA double-strand break sites following Dichloroacetate treatment (Fig. 6f). In addition, comet assays revealed increased residual DNA fragmentation in cells treated with Dichloroacetate and radiation (Fig. 6g). To further determine the in vitro activity of the triple-combination strategy across different EGFR backgrounds, clonogenic survival assays were performed using graded radiation doses (0, 2, 4, and 6 Gy) in EGFR T790M-mutant H1975 cells and EGFR-wild-type A549 cells. In H1975 cells, the combination of Osimertinib, Dichloroacetate, and radiation produced the most pronounced downward shift of the survival curve compared with double-combination treatment, indicating a strong dose-dependent radiosensitizing effect. By contrast, in A549 cells, the survival curves of the four treatment groups remained largely overlapping, and no substantial additional reduction in clonogenic survival was observed with the triple combination. These findings indicate that the in vitro activity of the triple-combination strategy is markedly stronger in EGFR T790M-mutant cells than in EGFR-wild-type cells (Fig. 6h). In vivo validation was performed using an EGFR T790M patient-derived xenograft (PDX) model. Dichloroacetate combined with radiation significantly suppressed tumor growth and delayed tumor progression compared with radiation alone. Furthermore, a triple-combination regimen incorporating an EGFR-EGFR T790M inhibitor, Dichloroacetate, and radiation provided additional therapeutic benefit, while the dual-combination regimen retained substantial antitumor activity (Fig. 6i). Finally, systemic toxicity was evaluated to assess treatment tolerability. Serum biochemical analyses showed no significant alterations in ALT, AST, BUN, or CK levels across treatment groups, indicating that the combination strategies exhibited acceptable safety profiles in vivo (Fig. 6j).

Fig. 6.

Triple targeting of EGFR T790M and lactate metabolism synergizes with radiotherapy: A mechanism-based proof-of-concept strategy. (a) Mechanism-guided drug repurposing screen to identify FDA-approved compounds targeting LDHA. Candidate drugs were evaluated in combination with radiation and Osimertinib, and their effects were assessed in cell-based assays and organoid models. The top six candidate compounds are shown. (b) Molecular docking and molecular dynamics (MD) simulation analysis of Dichloroacetate binding to LDHA. Representative binding conformations and stability of the Dichloroacetate-LDHA complex during MD simulation are shown. (c) Measurement of intracellular lactate concentrations in H1975 and PC9-EGFR T790M cells treated with radiation alone or radiation in combination with Dichloroacetate. (d) Extracellular acidification rate (ECAR) assessed by Seahorse analysis in H1975 and PC9-EGFR T790M cells treated with radiation alone or radiation combined with Dichloroacetate. (e) Analysis of DNA damage repair kinetics by γH2AX foci dynamics in H1975 and PC9-EGFR T790M cells treated with radiation alone or radiation combined with Dichloroacetate.Quantification was performed on at least 50 cells per condition. (f) Immunofluorescence co-localization analysis of RAD51 and γH2AX in H1975 and PC9-EGFR T790M cells under radiation alone or radiation combined with Dichloroacetate.Quantification was performed on at least 50 cells per condition. (g) Comet assays assessing residual DNA damage in H1975 and PC9-EGFR T790M cells following radiation alone or radiation combined with Dichloroacetate.Quantification was performed on at least 50 cells per condition. (h) Clonogenic survival assay of H1975 and A549 cells treated with radiation alone, radiation plus Osimertinib, radiation plus Dichloroacetate, or triple-combination treatment across graded radiation doses (0, 2, 4, and 6 Gy). (i) In vivo evaluation of radiation response in an EGFR T790M patient-derived xenograft (PDX) model. Tumor growth was monitored in mice treated with vehicle, radiation alone, EGFR-EGFR T790M inhibitor combined with radiation, Dichloroacetate combined with radiation, or triple-combination therapy. (j) Assessment of systemic toxicity and tolerability in vivo. Serum biochemical parameters, including ALT, AST, BUN, and CK, were measured in mice from the indicated treatment groups. Unless otherwise indicated, data are presented as mean ± SD from at least three independent biological replicates, with n = 3 or n = 6 depending on the specific assay. Technical duplicate or triplicate measurements were averaged within each biological replicate before statistical analysis. For immunofluorescence, PLA, comet assays, and organoid analyses, at least 50 cells or 50 organoids were quantified per condition in each independent biological replicate

Discussion

Radiotherapy constitutes a central component of local control and disease management in non-small cell lung cancer (NSCLC); however, its therapeutic efficacy varies substantially across distinct molecular subtypes [27]. In EGFR-mutant lung cancer in particular, although radiotherapy provides clear benefits in controlling local lesions [27], intrinsic and acquired tumor tolerance to radiation remains a major obstacle to durable treatment efficacy [28]. In recent years, accumulating evidence from multiple retrospective clinical cohorts and exploratory prospective studies has demonstrated that the combination of EGFR tyrosine kinase inhibitors (EGFR-TKIs) with radiotherapy can improve local control rates and prolong progression-free survival in a subset of patients with EGFR-mutant NSCLC [12, 29, 30]. Notably, this clinical benefit has also been reported in patient populations harboring acquired resistance mutations, including EGFR T790M [31, 32], suggesting that EGFR signaling may be deeply involved in the regulation of radiotherapy responses. Nevertheless, the molecular basis underlying this clinical synergy has remained largely undefined.

Previous mechanistic studies investigating the interaction between EGFR signaling and radiotherapy have primarily focused on the direct regulation of DNA damage repair by EGFR [33–35]. Extensive preclinical evidence indicates that ionizing radiation induces nuclear translocation of EGFR, where it forms complexes with key DNA repair proteins such as DNA-PKcs and Ku70/80, thereby enhancing the efficiency of non-homologous end joining (NHEJ). In parallel, activation of downstream PI3K/AKT and MAPK pathways following irradiation promotes cell survival and supports DNA repair processes [33, 36, 37]. Accordingly, pharmacological inhibition of EGFR disrupts these mechanisms, prolongs γH2AX persistence, and enhances radiation-induced cytotoxicity [38]. Together, these findings provide important molecular rationales for the radiosensitizing potential of EGFR-TKIs.

However, these models exhibit notable limitations. First, by concentrating primarily on direct interactions between EGFR and DNA repair proteins, they fail to explain the molecular context-dependence observed in clinical responses to EGFR-TKI plus radiotherapy. Second, EGFR has largely been treated as an isolated signaling node, with insufficient consideration of how radiotherapy, as a systemic cellular stress, reshapes the metabolic landscape of tumor cells. Most importantly, under the clinically prevalent acquired resistance background of EGFR T790M, the regulatory logic governing EGFR-radiotherapy interactions remains poorly characterized.

In recent years, tumor metabolic reprogramming has emerged as a key mechanism by which cancer cells adapt to therapeutic stress [39]. EGFR-mutant lung cancers are well recognized to exhibit enhanced glycolytic flux and increased lactate production [18]. In the context of radiotherapy, however, lactate has traditionally been regarded as a metabolic byproduct or a contributor to microenvironmental acidification, and whether it functions as an intracellular signaling molecule directly involved in DNA damage responses has remained largely unexplored. The discovery of histone lactylation has provided a novel epigenetic framework for lactate-mediated regulation, yet its role in radiotherapy responses-particularly in EGFR-mutant lung cancer-remains unknown.

Previous studies have shown that ionizing radiation can promote glycolytic reprogramming and lactate accumulation in tumor cells [40], indicating that radiation-associated lactate elevation itself is not unique to our model. In parallel, hypoxia is a well-established regulator of tumor glycolysis and lactate metabolism and may further shape radiation-associated metabolic adaptation through pathways such as HIF-1α-dependent transcriptional reprogramming [41]. Therefore, the metabolic effects of radiotherapy should be interpreted within a broader framework in which radiation, oncogenic signaling, and microenvironmental hypoxia may collectively influence tumor metabolic responses. In addition, the identification of histone lactylation as a metabolically responsive chromatin modification has provided a new conceptual link between intracellular lactate accumulation and epigenetic regulation [42]. Within this context, our findings extend the existing literature by showing that, in EGFR T790M-mutant NSCLC, radiation-amplified intracellular lactate is not merely a byproduct of stress metabolism, but is functionally coupled to histone lactylation and the establishment of a repair-permissive chromatin state that facilitates DNA damage repair and radiotherapy tolerance.

Against this background, the present study focuses on EGFR T790M, a clinically highly relevant and biologically distinctive mutation, to systematically dissect the intrinsic mechanisms driving radiotherapy tolerance. Our results demonstrate that EGFR T790M not only sustains tumor cell dependence on EGFR signaling but also, upon irradiation, triggers a specific metabolic adaptation program. Rather than acting solely as a DNA-damaging stimulus, radiotherapy cooperates with EGFR T790M signaling to markedly amplify glycolytic flux, leading to rapid intracellular lactate accumulation.

Mechanistically, we identify lactate dehydrogenase A (LDHA) as a critical downstream metabolic node that is highly responsive to radiotherapy in the EGFR T790M context. Irradiation promotes spatial proximity between EGFR T790M and LDHA and induces tyrosine phosphorylation of LDHA, thereby enhancing its enzymatic activity and sustaining lactate production. This finding redefines LDHA from a conventional “metabolic executor” to a functional effector molecule co-regulated by oncogenic signaling and radiotherapy.

Furthermore, we show that radiation-induced lactate is neither rapidly exported nor oxidized within mitochondria but instead accumulates intracellularly to participate in epigenetic regulation. Intracellular lactate drives genome-wide redistribution of histone H3 lysine lactylation (H3Kla), with selective enrichment at the promoters of DNA repair genes such as PRKDC and RAD51. Notably, this epigenetic modification induces only modest transcriptional changes, suggesting that its primary function is to establish a chromatin “poised” state that facilitates rapid DNA repair rather than to trigger strong transcriptional activation. Functionally, this epigenetic priming markedly enhances repair complex assembly, recruitment of repair proteins to DNA break sites, and the kinetics of DNA damage clearance.

From a translational perspective, using a mechanism-guided drug repurposing strategy, we identify the FDA-approved drug Dichloroacetate as a direct LDHA-targeting agent. In both in vitro and in vivo models, Dichloroacetate effectively suppresses radiation-induced lactate accumulation and DNA repair capacity, thereby significantly enhancing radiotherapy efficacy. These findings not only provide a mechanistic explanation for the clinical observations of EGFR-TKI and radiotherapy synergy but also suggest that further blockade of the metabolic-epigenetic adaptation axis may overcome EGFR T790M-associated radiotherapy tolerance.

From a broader translational perspective, the therapeutic value of our study lies not only in identifying a metabolic-epigenetic mechanism of radiotherapy resistance, but also in proposing a clinically approachable intervention strategy. Importantly, we did not begin with a predefined LDHA inhibitor. Instead, after identifying the EGFR T790M-LDHA-lactate-DNA repair axis as a key mechanism, we performed a screen of FDA-approved compounds and identified dichloroacetate (DCA) as a candidate agent capable of suppressing lactate-associated metabolic adaptation. This design highlights a drug repurposing strategy, namely, the reuse of an old drug in a new mechanistic setting. Compared with de novo drug development, such an approach may have greater translational potential because the pharmacologic properties and safety profiles of these agents are already partly understood [43]. At the same time, it should be emphasized that DCA is not a highly selective direct LDHA inhibitor [44]. Rather, in the context of our study, it should be viewed as a proof-of-concept pharmacologic tool showing that pharmacologic suppression of lactate-associated metabolic adaptation can enhance radiotherapy response. In addition, current clinical evidence indicates that no selective LDH/LDHA-targeted therapy has yet been approved for cancer treatment, and the clinical development of this class remains limited. A phase II study of DCA in advanced non-small cell lung cancer also suggested that this type of metabolic intervention may be more valuable when integrated with established treatment modalities, rather than used as monotherapy [45].

In current clinical practice, treatment of EGFR-mutant NSCLC remains centered on EGFR-targeted therapy [46]. Osimertinib has long served as a standard first-line backbone, while newer intensified regimens, such as amivantamab plus lazertinib, further illustrate that rational combination strategies can improve outcomes beyond EGFR-TKI monotherapy [47]. In parallel, radiotherapy and other local ablative approaches are increasingly incorporated into the management of residual lesions, oligometastatic disease, and oligoprogressive disease during targeted treatment [5, 48]. Within this context, our findings do not suggest that LDH/lactate targeting should replace current standards of care. Rather, they support the concept that metabolic intervention may serve as a rational combination strategy to improve the efficacy of existing therapeutic backbones. In particular, for tumors in which radiotherapy induces a lactate-dependent DNA repair program, combining radiotherapy and EGFR inhibition with metabolism-targeting agents may represent a promising mechanism-guided direction for future translational development.

Integrating the experimental evidence from Figs. 1, 2, 3, 4, 5 and 6, we propose a unified model (Fig. 7): in the EGFR T790M background, irradiation activates EGFR signaling, which enhances EGFR-LDHA interaction and promotes LDHA tyrosine phosphorylation, thereby driving glycolytic enhancement and intracellular lactate accumulation. Lactate acts as a signaling metabolite that enters the nucleus and induces H3Kla, establishing an epigenetically primed state at DNA repair gene regulatory regions, thereby accelerating DNA repair and promoting cell survival. EGFR inhibition or LDHA-targeted intervention disrupts this pathway at distinct levels, ultimately restoring radiosensitivity.

Fig. 7.

Proposed model illustrating how radiation-induced EGFR activation promotes radiotherapy resistance through an LDHA-lactate-epigenetic repair axis in EGFR T790M-mutant lung cancer. Schematic model summarizing the working mechanism identified in this study. Radiation activates EGFR signaling in EGFR T790M-mutant lung cancer cells and enhances EGFR-dependent tyrosine phosphorylation of LDHA, thereby increasing LDHA activity and amplifying glycolytic reprogramming. This results in persistent intracellular lactate accumulation. The accumulated lactate is diverted to histone H3 lysine lactylation (H3Kla), which becomes enriched at the promoter regions of DNA repair genes such as PRKDC and RAD51, generating a repair-permissive chromatin environment. As a consequence, recruitment of DNA repair factors to DNA double-strand breaks is facilitated, γH2AX foci are resolved more efficiently, and radiotherapy resistance is promoted. Therapeutically, inhibition of EGFR signaling and/or blockade of the LDHA-lactate axis disrupts this pathway and restores radiosensitivity

Importantly, EGFR T790M should not be regarded merely as a passive marker of resistance to targeted therapy, but rather as a functional mutation capable of systemically reshaping tumor radiotherapy responses. By centering on EGFR T790M, this study reveals a metabolic-epigenetic-repair axis that underlies radiotherapy tolerance, offering new insights into the radiobiology of EGFR-mutant lung cancer and establishing a theoretical foundation for mechanism-based combination radiotherapy strategies. Although our data support a particularly prominent role for this radiation-induced lactate-epigenetic repair program in the EGFR T790M setting, we cannot exclude the possibility that related metabolic-epigenetic remodeling may also occur in other EGFR-mutant contexts. Defining the degree of shared versus mutation-specific regulation will require further investigation in broader EGFR-mutant models.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Hongxia Cheng, Lu Meng, Shilan Luo, Chenhui Zhao, Xiesong Luo, Jingdan Pang, Qingyu Jiang, Yanghai Xiong, and Peiwen Zhu generated the data including supplementary data; Qingyu Jiang collected and provided the clinical samples; Chenhui Zhao, Jingdan Pang, and Yanghai Xiong performed bioinformatics and statistics analysis; Hongxia Cheng, Lu Meng drafted the manuscript; all authors read and revised the manuscript; Kun Guo and Xiaomei Gong guided the project.

Funding

National Natural Science Foundation of China (82473378), the Shanghai Pulmonary Hospital Research-Oriented Physician Talent Program (LYRC202407), Tongji University Medicine-X Interdisciplinary Research Initiative (TJ-FK-YXJC022)and Shanghai Municipal Health Commission Seed Funding Program for Research and Translation of New Medical Technologies (2025ZZ2011).

Data availability

No datasets were generated or analysed during the current study.

Code availability

Not applicable.

Declarations

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.