Putrescine functions as a metabolic checkpoint in replication stress-induced senescence

Abstract

Cellular senescence represents a fundamental biological response to replication stress and other genotoxic insults, acting as both a barrier to malignant transformation and a driver of age-related tissue dysfunction. Here, we dissect the metabolic remodeling that accompanies senescence triggered by sustained activation of the replication licensing factor CDC6. Induction of CDC6 in human bronchial epithelial cells provoked a biphasic response characterized by transient hyperproliferation followed by accumulation of senescence-associated β-galactosidase-positive cells. Targeted metabolomics revealed an early increase in intracellular putrescine that was followed by pronounced decline at senescence onset. Functional studies demonstrated that putrescine supplementation attenuated CDC6-induced senescence, whereas knockdown of ODC1, the rate-limiting enzyme in putrescine biosynthesis, accelerated it and increased TP53 accumulation. Mechanistically, CDC6 controls the ODC1–putrescine axis through ERK and GSK3β-mediated regulation of MYC, whereby early-phase ERK signaling stabilizes MYC to enhance polyamine biosynthesis, while prolonged CDC6 activation triggers GSK3β-dependent MYC degradation, ODC1 downregulation, and commitment to senescence. Targeted re-analysis of publicly available single-cell RNA-sequencing datasets from COVID pneumonia patients revealed elevated CDC6 expression alongside reduced MYC and ODC1 levels in alveolar epithelial cells exhibiting markers of senescence. Collectively, these findings identify putrescine as a metabolic checkpoint in replication stress-induced senescence and reveal a MYC-orchestrated signaling–metabolic circuit that temporally integrates oncogene activation with cell fate decisions.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-025-05962-9.

Introduction

Initially identified by Hayflick and Moorhead [1], senescence is recognized as a major tumor-suppressive mechanism as well as contributor to tissue homeostasis, aging and various age-related pathologies [2, 3]. Typically, it is manifested by stable cell cycle arrest, increased activity of senescence-associated β-galactosidase (SA-β-Gal), and the formation of senescence-associated heterochromatin foci (SAHF) [2, 4]. Senescent cells also secrete a mixture of inflammatory, mitogenic and fibrogenic soluble factors, metabolites and extracellular vesicles, collectively known as the senescence-associated secretory phenotype (SASP), which can reinforce the senescence state, modify the local tissue microenvironment, and impact aging [2, 4, 5]. The composition and intensity of SASP may vary according to cell type and the initiating stressor. For example, a core set of proinflammatory SASP components, including interleukin-6 (IL-6), interleukin-8 (IL-8), and matrix metalloproteinase 3 (MMP3), is secreted in response to telomere shortening, DNA damage, and oncogene activation [5] but is absent in cells ectopically expressing p16^INK4a or p21 [6].

Senescence can also be triggered by replication stress resulting from various endogenous and exogenous factors such as oncogene activation, metabolic disturbances, ionizing radiation, and chemotherapeutic agents, which compromise replication fork stability and impair genome duplication fidelity [7–10]. Among these factors, overexpression of the replication licensing factor CDC6 is a well-established inducer of replication stress, capable of promoting either proliferation and oncogenic transformation or senescence, depending on the cellular context and the functional status of tumor suppressor pathways [9]. Thus, in cells with functional TP53 and p16^INK4a/Rb pathways, including human bronchial epithelial cells (HBECs) [11, 12], CDC6 overexpression deregulates normal DNA synthesis, leading to activation of a DNA damage response driving senescence. However, when these pathways are inactivated, CDC6 facilitates uncontrolled proliferation and epithelial-to-mesenchymal transition, promoting malignant transformation [9, 11–13]. Indeed, elevated levels of CDC6 have been reported in numerous human malignancies [14–16].

Despite loss of proliferative capacity, senescent cells remain alive and metabolically active [17]. The metabolic reprogramming of cells undergoing senescence is not fully understood. Available data indicate senescence-associated changes in glycolysis, gluconeogenesis, and pentose phosphate metabolism in a cell type-dependent manner [17]. Thus, replicative senescence in myoblasts is associated with impaired glycolysis [18], whereas in human fibroblasts and keratinocytes is associated with increased glycolysis, evidenced by elevated intracellular pyruvate levels [18–21]. The identification of intracellular NAD⁺/NADH as a key regulator of pro-inflammatory SASP intensity [22] further underscores critical links between cellular metabolism and hallmarks of senescence and emphasizes the need for further investigations into the influence of metabolic pathways on senescence and their functional output.

Εmerging evidence in model organisms and humans also points to an important role of polyamine metabolism in aging. The arginine–ornithine–polyamine pathway involves the conversion of arginine to ornithine by arginase, followed by decarboxylation of ornithine to putrescine by ornithine decarboxylase 1 (ODC1), the rate-limiting enzyme in this process [23, 24]. In turn, putrescine serves as a precursor for the biosynthesis of spermidine, a higher-order polyamine that delays aging in model systems by promoting autophagy and conferring anti-inflammatory properties (reviewed in [23]). While putrescine has been extensively studied in plants, where it mitigates biotic and abiotic stress and inhibits senescence [25], its role in senescence in mammalian cells remains widely unknown.

In this study, we aimed to characterize the metabolic changes associated with replication stress-induced senescence, using CDC6 overexpression in HBECs as a relevant model. Through targeted metabolomic profiling, we sought to identify intracellular and extracellular metabolites affecting entry to senescence, and to assess their functional roles in modulating this process. Among several candidates, we describe putrescine as a novel metabolic suppressor of replication stress-induced senescence and report a MYC-ODC1-putrescine axis serving as rheostat that controls replication stress-induced cell fate determination.

Materials and methods

Chemicals and reagents

Acetonitrile (ACN) and methanol (MeOH) were LC-MS grade and purchased from Sigma-Aldrich (St. Louis, USA). Pure water (H₂O) (18.2 MΩ cm-1) was acquired in a Milli-Q device (Millipore Purification System, Merck Darmstadt, Germany). Ammonium formate (HCOONH₄) was obtained from Chem-Lab (Zedelgem, Belgium) and formic acid (F.A., >98%) was purchased from Riedel-de Haën^® (Sigma–Aldrich, Steinheim Germany).

Cell lines

The development of Tet system for doxycycline-inducible expression of CDC6 in human bronchial epithelial cell (HBEC) has been previously reported [11]. The cell line was cultured in keratinocyte serum free medium (Gibco #17005042) supplemented with 50 µg/ml (BPE) and 5 ng/ml human recombinant Epidermal Growth Factor (EGF) and maintained at 37 °C and 5% CO2. According to the manufacturer, the keratinocyte medium contains 0.19 mg/L putrescine.

For CDC6 induction, cells were cultured in the presence of doxycycline (Dox) (Sigma Aldrich #D9891-1G) at a final concentration of 1 µg/mL. Dox was replenished every second day. Untreated cells (Tet OFF) served as controls for all experiments. To inhibit MEK1/2 pathway, PD98059 (Thermo Fisher Scientific #PHZ1164) was added to medium at the concentration of 10 µM. To inhibit the proteasomal activity, MG132 (Thermo Fisher Scientific #J63250.MCR) was added to medium at the concentration of 5µΜ. For the induction of drug-induced senescence, cells were cultured in the presence of 5µΜ etoposide.

Evaluation of cell proliferation

Approximately 50,000 cells were seeded per well in 12-well plates, treated with DOX, and harvested by trypsinization. Cell numbers were determined by counting with a hemocytometer after resuspension in PBS containing trypan blue.

SA-β-galactosidase cell staining

Senescence-associated-β-galactosidase staining was performed according to Debacq-Chainiaux’s protocol [26]. The cells were seeded and cultured in 12-well plates. After DOX treatment, the cells were washed with 1x PBS, fixed for 4 min with 4% formaldehyde in 1x PBS and washed 3 times with 1x PBS. Next, staining solution (20 mg/ml X-gal in dimethylformamide, 0.2 M citric acid/Na phosphate buffer (pH = 6.0), 100 mM potassium ferrocyanide, 100 mM potassium ferricyanide, 5 M NaCl and 1 M MgCl₂) was added (1–2 ml per 35 mm dish), and the mixture was incubated overnight at 37 °C in a dry incubator (without CO₂). After overnight incubation, the cells were observed under a Zeiss Axiovert 40 C inverted phase contrast microscope with a 10x objective (100x magnification).

Quantitative SA-β-galactosidase assay

The mammalian beta-Galactosidase Assay Kit (Thermo Fisher Scientific #75707) was used for senescence quantification, following the manufacturer’s protocol. Cells were harvested, lysed with M-PER reagent, and centrifuged at 27,000 × g for 15 min. Cell extracts (50 µl) were incubated with the assay reagent for 30 min at 37 °C. Absorbance was measured at 405 nm after stopping the reaction.

Cell cycle analysis by flow cytometry

For cell cycle analysis, 500,000 cells were seeded in 6-well plates and treated with Dox. Cells were harvested, fixed with 80% ethanol, and stained with 50 µg/mL propidium iodide (PI) and 10 µg/mL RNase A (Invitrogen #AM2271) diluted in staining buffer (100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM CaCl₂, 0.5 mM MgCl₂, and 0.1% NP-40). DNA content was analyzed using a FACSCalibur flow cytometer (BD Biosciences).

Annexin V/propidium iodide (PI) staining assay

Apoptosis was assessed using an Annexin V-FITC apoptosis detection kit (Cayman Chemical #601420). Cells were treated with DOX, harvested, and stained with Annexin V-FITC and PI in Annexin V Binding Buffer, following the manufacturer’s instructions. Samples were analyzed using a FACSCalibur flow cytometer (BD Biosciences).

RNA interference

siRNA-mediated knockdown was performed as previously described [27, 28]. ODC1 (Invitrogen #4390824) and luciferase siRNAs (Ambion #AM16204, negative control) were transfected using Lipofectamine RNAiMax (Invitrogen #56531).

Protein extraction and western blot analysis

Protein extraction, quantification and electrophoresis were performed as previously described [27]. For immunoblotting, the following primary antibodies were used: CDC6 (dilution 1:000, Cell Signaling #3387S), p53 (dilution 1:500, Santa Cruz #47698), MYC (dilution 1:1000, Cell Signaling #5605), ODC1 (dilution 1:1000, Cell Signaling #52238), phospho-GSK3β (dilution 1:1000, Sigma Aldrich #SAB4300237), phospho-ERK1/2 (dilution, 1:000 Cell Signaling #4370), ERK1/2 (dilution 1:1000, Cell Signaling #9102), phospho-MYC S62 (dilution 1:500, CUSABIO #PA000729), phospho-MYC T58 (dilution 1:500, CUSABIO #PA000503), phospho-P65 NF-κB (dilution 1:1000, Santa Cruz #136548), P65 NF-κB (dilution 1:1000, Santa Cruz #8808), phospho-MEK1/2 (dilution 1:1000, Cell Signaling #9121), MEK1/2 (dilution 1:1000, Cell Signaling #9122), phospho-Histone H2Ax (dilution 1:500, Cell Signaling #9718) and actin (ACTB; dilution 1:500, Sigma Aldrich #MAB1501). Anti-mouse (Sigma Aldrich #A9044) and anti-rabbit (Cell Signaling #7074S) HRP-linked secondary antibodies were used at dilutions of 1:4000 and 1:3000.

RNA extraction, cDNA preparation and qRT‒PCR

RNA from cells was isolated via Nucleospin RNA (Macherey–Nagel #740955) according to the manufacturer’s protocol. cDNA synthesis was carried out via a high-capacity cDNA reverse transcription kit (Applied Biosystems #4368814). Quantitative real-time PCR was performed according to the TaqMan™ Fast Universal PCR Master Mix (2X) protocol (Applied Biosystems # 4352042), as previously described [27]. The following TaqMan human probes were used: MYC (Thermo Fisher, Hs00153408_m1), ODC1 (Thermo Fisher, Hs00159739_m1) and ACTB (Thermo Fisher, Hs99999903_m1).

Ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS)

For the metabolomics-based analysis, H₂O: MeOH 1:1 (v/v) was added into the cell pellets in a 3:1 Vs/Wp (solvent volume/pellet weight) ratio. Then, 100 µL of extract were evaporated to dryness under vacuum. The residue was reconstituted with 60 µL of ACN: H₂O 95:5 v/v and the clear supernatant subjected to targeted HILIC UHPLC-MS/MS analysis. Regarding the cell culture media, 50 µL of the cell supernatant were diluted with 150 µL of ice cold ACN prior to analysis.

For the targeted analysis of putrescine in cell culture, H₂O: ACN, 1:1, v/v was added into the pellets at a solvent volume-to-pellet weight (Vs/Wp) ratio of 1.6:1. The sample was vortex-mixed for 5 min and centrifugated for 10 min at 10,000 rpm, 4^◦C. One hundred microliters of the clear supernatant were transferred to a LC-MS glass vial for analysis.

Metabolomics-based analysis was performed in an ACQUITY UPLC H-Class chromatography system coupled to a Xevo TQD mass spectrometer (Waters Corporation, Milford, USA), operating in ESI positive and negative ionization mode. Chromatographic separation was performed in an Acquity BEH Amide column (2.1 mm × 150 mm, 1.7 μm) with an Acquity UPLC Van-Guard pre-column (Waters, UK) of the same packing material. A binary mobile phase system was used consisted of (A) ACN: H₂O 95:5 v/v and (B) ACN: H₂O 30:70 v/v both containing 10 mM HCOONH₄, while the gradient program was as followed: 0–4 min 100% Α, 4–25 min 60% A, 25–30 min 15% A and 30–40 min 100% Α.

Detection was performed using multiple reaction monitoring (MRM) mode, where cone voltage and collision energy were optimized separately for each compound after direct infusion into the MS. Capillary voltage was set at + 3.5 kV for positive and − 3.5 kV for negative ionization mode. Desolvation temperature, desolvation flow and cone gas flow were set at 350 °C, 650 L/h and 50 L/h, respectively.

Putrescine analysis was performed under similar chromatographic conditions. The analysis was performed in an ACQUITY UPLC H-Class chromatography system coupled to SCIEX 6500 + mass spectrometer (AB Sciex Pte. Ltd, USA) in positive ESI ionization mode. Chromatographic separation was performed using the same analytical column and pre-column. Mobile phase used consisted of (A) ACN: H₂O 95:5 v/v and (B) ACN: H₂O 30:70 v/v, both containing 10 mM HCOONH₄ and formic acid (pH 3). The gradient program was as followed: 0–4 min 70% Α to 30% A, followed by 4 min of column equilibration. Detection was performed using again MRM mode, where cone voltage and collision energy were optimized for both putrescine and its internal standard (IS).

A Quality Control sample (QC) for each matrix was used throughout the analytical batch in order to assess system’s repeatability and analytical performance [29, 30]. QC sample was prepared by mixing equal volumes of all tested real samples. QC was analyzed 5 times at the beginning of the analytical batch, for system equilibration and then every other 5 samples.

Single-cell analysis of transcriptional landscape of human alveolar epithelial cells

Publicly available gene expression matrices were obtained for the analysis of healthy donors (HD; EGAS00001004344; [31]) or patients with pneumonia upon viral (EGAS00001004717; [32]), viral coupled with fungal (EGAD00001011342; [33]) or bacterial (GSE155249; [34]) infection. Initial quality control was performed on each dataset independently. Data were pre-processed using CellBender (0.3.0) for detection of ambient RNA contamination traces. Additional exclusion criteria for low-quality cells included expression of less than 200 genes and/or 1,000 unique sequenced reads per cell, distribution of greater than 10% or 30% of total counts in mitochondrial or ribosomal genes, respectively, alongside doublet detection (DoubletFinder, v2.0.4). A total of 7,514 sc transcriptomes were obtained following quality control and filtering and processed for subsequent data normalization, feature selection, integration and dimensionality reduction using the Seurat R package (v5.2.1) for the identification of 4,000 highly variable genes. Unsupervised cluster identification was performed employing nearest neighbour analysis. Orthogonal strategies were employed for cluster annotation including manual marker gene inspection, semi-automatic annotation and label transfer from published reference datasets. Uniform manifold approximation and projection (UMAP) was implemented for the generation of two-dimensional representations (n.neighbors = 30, min.dist = 0.5, metric = cosine). Alveolar epithelial cell clusters were selected for the projection of gene expression scores on the UMAP plots constructed. The senescence gene expression signature was based on the following markers: IL1B, IL6, IL8, CDKN2A, CDKN1A, LMNB1, CCL5, MMP3, SERPINE1, CXCL10, CCL2, TNF, GLB1.

Data handling and statistics

Collected metabolomic data were processed using MassLynx^® and TargetLynx^® V4.1 software (Waters, Milford, MA, USA) and SCIEX OS software (AB Sciex Pte. Ltd, Framingham, MA, USA) then further evaluated using different statistical analysis tools. Metabolites assessed were either present in more than 80% of real samples in each group and their coefficient variation (CV) in the QC sample was less than 0.3%. Univariate statistical analysis was performed using Microsoft Excel (Microsoft corporation, USA) and GraphPad Prism 8 (GraphPad Prism v.8.0 for Windows, GraphPad Software, Boston, Massachusetts USA). Two-tailed t-test with unequal variance or one-way ANOVA, followed by post-hoc analysis, false discovery rate (FDR) correction, setting a p-value threshold at 0.05 and log₂FoldChange were performed. For multivariate statistical analysis, Principal Component Analysis (PCA) and Orthogonal Partial-Least Squares-Discriminant Analysis (OPLS-DA) in UV scaling were performed using SIMCA P + 13.0 software (Umetrics, Malmö, Sweden). Validity of constructed models was based on their R2X, R2Y and Q2Y values and p-value of CV-ANOVA (Cross validation, R2X: fraction of the variation in X explained by the model, R2Y: total sum of variation in Y explained by the model and Q2Y: goodness of prediction). Different visualization plots were also addressed for the evaluation of possible biomarkers.

Statistical analyses of collected qPCR, protein, SA-β-galactosidase, Annexin V-FITC/PI staining and flow cytometry data were performed using GraphPad Prism 8 (GraphPad Prism v.8.0 for Windows, GraphPad Software, Boston, Massachusetts USA). One-way, two-way ANOVA and unpaired t-test were used, as appropriate, to assess statistical significance across groups. Protein expression levels were semi-quantified with ImageJ (https://imagej.net/ij/). All observations reported in this manuscript were based on ≥ 3 biological replicates. Results are presented as mean ±, standard deviation (SD) and p-value, where ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

Results

Temporal effects of CDC6 over-expression on senescence

Previous studies have demonstrated that CDC6 overexpression in fibroblasts and epithelial cells triggers morphological and molecular characteristics of senescence [11, 12]. Here, we analyzed in more detail the temporal effects of CDC6 overexpression on cell cycle and onset of senescence using HBECs stably transfected with a doxycycline (DOX)/tetracycline-inducible CDC6 transgene (HBEC^Tet−CDC6) [11].

As shown in Fig. 1 A and Supplementary Fig. 1 A, addition of DOX to HBEC^Tet−CDC6 cultures (Tet ON) resulted in a time-dependent increase in CDC6 expression. This was associated with elevated cell proliferation during the first two days of DOX treatment compared to Tet OFF controls, but progressively decreased cell numbers thereafter (Fig. 1B). These changes aligned with the effects of CDC6 on cell cycle with higher HBEC numbers in the S and G2/M phases up to two days of CDC6 induction, and an ensuing progressive decline by day 6 (Fig. 1 C). The pronounced reduction in cell proliferation on days 3 and 6 was accompanied by approximately 11% and 15% cell death, respectively (Fig. 1 C; see also Suppl. Figure 2), which cannot fully explain the significant decrease in cell numbers observed at these time points.

Fig. 1.

CDC6 over-expression induces senescence. (A) Immunoblot showing time-dependent induction of CDC6 in HBEC-Tet^CDC6 cells following addition of doxycycline (Tet ON). The housekeeping β-actin (ACTB) serves as loading control. (B) Cell counts of HBEC^Tet−CDC6 cultures (absolute cell numbers) induced to express CDC6 (Tet ON) versus uninduced (Tet OFF) controls. Note the temporal increase in proliferation of Tet ON cells up to day 2 and the reduction in proliferation thereafter. (C) Distribution of HBEC^Tet−CDC6 cultures to different cell cycle phases following addition of doxycycline (Tet ON) for various time intervals compared with Tet OFF cells. (D) Induction of senescence following CDC6 overexpression. (i) Representative SA-β-Galactosidase staining (scale bar: 20 μm) and (ii) quantitative assessment of SA-β-Galactosidase activity of HBEC^Tet−CDC6 cultures treated with doxycycline (Tet ON) for various time intervals, compared to Tet OFF cultures. Data shown is the mean ± SD of a minimum of n = 3 independent experiments. Statistical analysis was performed using one-way ANOVA and Dunnett’s test. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001 and ^****p < 0.0001; d; day

Senescence-associated β-galactosidase (SA-β-gal) staining (Fig. 1Di) and quantitative assessment of SA-β-gal activity (Fig. 1Dii) were performed and demonstrated a modest 2-fold senescence induction on day 2, compared to Tet OFF controls, elevated to 3.8-fold by day 3 and to 6.4-fold by day 6 of DOX treatment.

Together, these results demonstrate that CDC6 over-expression in non-transformed lung epithelial cells provokes a biphasic response characterized by transient hyperproliferation followed by cessation of cell cycle and gradual entry to senescence.

Metabolic changes associated with CDC6-induced senescence

We investigated the metabolic changes accompanying entry to senescence in CDC6-overexpressing HBECs. A previously validated targeted HILIC-MS/MS method [35] was applied in lysates and supernatants of HBEC^Tet−CDC6 cultures exposed to DOX for 6 days in the presence or absence of DOX, allowing for the semi-quantification of over 100 metabolites from various biochemical pathways, including the citric acid cycle, amino acid metabolism, and glycolysis. A total of 34 metabolites in the cell samples and 51 in the cell culture media met the established criteria (see Materials & Methods) and were further considered for statistical analysis.

Principal component analysis (PCA) was performed on all analyzed and quality control samples, demonstrating a clustering that underscores the robustness of the analytical performance (Suppl. Figure 2A1 and 2B1). Subsequently, orthogonal partial least squares discriminant analysis (OPLS-DA) models were generated for pairwise comparisons across the two sample matrices.

When comparing the intracellular metabolic profile of Tet OFF versus Tet ON groups, OPLS-DA scores plot model indicated a modest differentiation with values (R2X = 0.574, R2Y = 0.954, Q2Y = 0.795, and CV-ANOVA 0.06) (Suppl. Figure 2Α2). Five metabolites, namely adenosine, leucine, valine, proline and putrescine were significantly decreased in the 6 d Tet ON group (Fig. 2 A).

Fig. 2.

Replication stress caused by CDC6 over-expression induces changes in intracellular and extracellular metabolites. Box-plots of statistically significant changes in (A) intracellular and B) extracellular metabolites assessed by HILIC MS/MS. “Peak area” on y-axis denotes the intensity of the MS/MS signal. Data shown is the mean ± SD of n = 5 independent experiments; *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001

A strong statistical differentiation was observed in the extracellular metabolic profile when comparing the Tet OFF and Tet ON groups (Suppl. Figure 2B). The OPLS-DA score plot (t₁ vs. t₂) demonstrated clear separation between these groups (Suppl. Figure 2B2). As shown in Fig. 2B, several extracellular metabolites were significantly affected by CDC6, with elevated levels of acetylcarnitine, creatine, glucose, hypoxanthine, phenylalanine, and tyrosine but reduced lactate and proline detected in the Tet ON versus Tet OFF group (Fig. 2B). All significantly altered metabolites had VIP values > 1.0 and p-values < 0.05.

Glucose does not play a major role in CDC6-induced senescence

Previous studies have indicated that high glucose levels promote senescence in fibroblasts, endothelial and renal epithelial cells [36, 37] and, conversely, glucose restriction in culture medium extends replicative lifespan in human diploid fibroblasts [38]. Given that CDC6-induced senescence is associated with increased extracellular glucose levels (Fig. 2B), we posited that culture medium glucose might delay senescence onset. To test this, HBEC^Tet−CDC6 cells were cultured either in standard medium (SM) containing 1,042.27 mg/L glucose, low glucose medium (LG) containing 104.23 mg/L glucose, or glucose-free medium (NG), with or without DOX for 1 to 6 days. Quantification of SA-β-gal activity showed that extracellular glucose levels modulation did not impact CDC6 overexpression-induced senescence (Fig. 3).

Fig. 3.

Modulation of extracellular glucose levels does not impact senescence induced by CDC6 overexpression. Bar charts showing the fold change in SA-β-Galactosidase activity at day 1 (A), day 2 (B), day 3 (C) and day 6 (D) of doxycycline treatment (Tet ON) versus control cultures (Tet OFF) in standard culture (SM) medium containing 1042.27 mg/L glucose, low glucose (LG) medium supplemented with104.23 mg/L glucose or without glucose (NG). Data shown is the mean ± SD of n = 3 independent experiments. Statistical analysis was performed using one way-ANOVA and Tukey’s test. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001

Exogenous putrescine suppresses CDC6-induced senescence

Putrescine inhibits senescence in plants [25], and other components of the polyamine pathway have been implicated in promoting lifespan in various model organisms [23]. Given that intracellular putrescine levels decline by over 50% by day 6 of CDC6 induction, coinciding with the establishment of senescence (see Figs. 2A and 1D), we investigated whether exogenous putrescine supplementation could counteract CDC6-induced senescence in HBEC^Tet−CDC6 cells. Lung epithelial cells uptake exogenous putrescine [39] via mechanisms involving P-type ATPases and/or solute carriers (SLCs) [40].

HBEC^Tet−CDC6 cells were expanded in media containing standard (SM; 0.19 mg/L), semi-double (SD; 0.285 mg/L) or double (Hi; 0.38 mg/L) putrescine concentration, in the presence or absence of DOX for 1 to 6 days. Measurement of SA-β-gal activity revealed a concentration-dependent suppression of CDC6-induced senescence (Fig. 4). Thus, the Hi putrescine concentration ameliorated SA-β-gal activity on day 2 (Fig. 4B) and reduced senescence by more than 50% by day 6 of CDC6 induction (Fig. 4D).

Fig. 4.

Exogenous putrescine alleviates CDC6-induced senescence. Bar charts showing the fold changes in SA-β-Galactosidase activity at day 1 (A), day 2 (B), day 3 (C) and day 6 (D) of doxycycline treatment (Tet ON) versus control cultures (Tet OFF) in standard culture (SM) medium containing 0.19 mg/L putrescine, medium containing semidouble (SD; 0.285 mg/L) or double concentration of putrescine (high, Hi; 0.38 mg/L). Data shown is the mean ± SD of a minimum of n = 3 independent experiments. Statistical analysis was performed using one way-ANOVA and Tukey’s test. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001

Changes in endogenous putrescine and ODC1 expression levels following CDC6 induction

The above findings (Figs. 2A and 4) prompted further investigation of the time-dependent changes in intracellular putrescine levels following CDC6 induction, and the molecular underpinnings of putrescine regulation.

First, using a targeted LC-MS/MS method we observed a significant increase in endogenous putrescine levels on day 1 of DOX treatment and subsequent decline, falling below basal levels by day 6 (Fig. 5 A). Next, we focused on ornithine decarboxylase 1 (ODC1), the rate-limiting enzyme for the conversion of ornithine into putrescine [41] by determining the temporal changes in ODC1 expression upon induction of CDC6. In parallel, we examined the expression of MYC, which transcriptionally regulates ODC1 [42, 43], TP53 as a regulator of CDC6-induced senescence in HBECs [11] and γΗ2Αx as additional surrogate marker of DNA damage response. We found a gradual increase in both ODC1 and MYC levels up to day 1 of DOX treatment which declined by day 2, and fell below basal levels by day 3 of CDC6 induction (Fig. 5B and Suppl. Figure 1B). In contrast, TP53 and γΗ2Αx progressively accumulated following DOX treatment (Fig. 5B and Suppl. Figure 1B). Therefore, CDC6 overexpression results in a transient increase in both ODC1 and putrescine which decline below basal levels as cells progressively enter senescence.

Fig. 5.

Effects of putrescine on CDC6-induced TP53 and MYC signaling. (A) Changes in intracellular putrescine levels following induction of CDC6 assessed by a targeted LC-MS/MS method. Data shown is the mean ± SD of n = 3 independent experiments. (B) Entry to senescence following CDC6 overexpression is associated with accumulation of TP53 and γH2Ax and reduced expression of MYC and its target, ODC1. Panel Bi shows a representative (n = 3) immunoblot of CDC6, TP53, MYC, γH2ΑX and ODC1 expression in HBECTet−CDC6 cells following addition of doxycycline (Tet ON) versus uninduced, Tet OFF cells. ACTB (β-actin) serves as loading control. The mean (± SD) fold changes of ODC1 and MYC expression levels from 3 independent experiments is shown in panel Bii. Statistical analysis was performed using one way-ANOVA and Tukey’s test. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. (C) Addition of putrescine in the culture medium of HBECTet−CDC6 cells ameliorates the effects of CDC6 on TP53 and MYC pathways. Panel Ci shows a representative immunoblot (n = 3) of CDC6, TP53, MYC, ODC1 and ACTB levels in Tet ON versus Tet OFF HBECTet−CDC6 cells in the presence of different concentrations of putrescine: standard culture (SM) medium containing 0.19 mg/L putrescine, semi-double (SD; 0.285 mg/L) or double concentration of putrescine (high, Hi; 0.38 mg/L). The graph of Cii shows the mean (± SD) fold changes of MYC and ODC1 expression levels from 3 independent experiments. Statistical analysis was performed using one way-ANOVA and Sidak’s test. ∗p < 0.05, **p < 0.01 and ****p < 0.0001. (D) Histogram showing the mean ± SD fold change mRNA levels of MYC and ODC1 on day 3 of DOX treatment in the presence of different putrescine concentrations (n = 3), as described in (C) Statistical analysis was performed using one way-ANOVA and Sidak’s test. *p < 0.05, ***p < 0.001 and ****p < 0.0001

Given that putrescine suppresses CDC6-induced senescence (Fig. 4), we investigated its impact on the expression of ODC1, MYC, and TP53 during the early phase of senescence induction, focusing on day 3 of DOX treatment (see Fig. 1B and D). As illustrated in Fig. 5 C, high-dose (Hi) putrescine supplementation markedly reduced TP53 and restored the expression of both MYC and its downstream target ODC1 to near-control levels observed in Tet OFF cells. Intermediate (SD) concentrations had a more modest impact (Figs. 5C(i), 5C(ii) and Suppl. Figure 1 C). These effects were further validated at the RNA level by RT-qPCR, which confirmed the recovery of MYC and ODC1 expression following putrescine treatment (Fig. 5D).

To assess whether the MYC–ODC1 axis is implicated in other senescence triggers, HBECs were treated with 5 µM etoposide and analyzed over 4 days. SA-β-Gal activity increased progressively, while MYC and ODC1 protein levels declined markedly by day 3–4 (Suppl. Figure 4). Unlike CDC6, no early MYC/ODC1 upregulation was observed upon etoposide treatment, indicating context-dependent regulation of this metabolic axis (Suppl. Figure 4). Therefore, whereas degradation of MYC is a common feature across several senescence triggers (Suppl. Figure 4 and [44]), early MYC upregulation may be specific to CDC6-induced replication stress.

To examine whether cell cycle redistribution contributes to the metabolic and molecular changes observed upon CDC6 induction, we synchronized HBECs via serum starvation and re-stimulated them with or without CDC6 overexpression. We found that SA-β-Gal, MYC/ODC1 expression, and putrescine levels followed similar trajectories regardless of prior synchronization (Suppl. Figure 5), suggesting that CDC6-induced metabolic rewiring occurs independently of initial cell cycle phase. While these results argue against major confounding effects from heterogeneous cell cycle distribution, we cannot entirely exclude subtle contributions from specific phases. Future single-cell or cell cycle-resolved approaches will be instrumental in dissecting the temporal coordination between cell cycle state and the onset of senescence-associated metabolic remodeling.

Knockdown of ODC1 accelerates CDC6-induced senescence

On the basis of the aforementioned observations, we investigated whether ODC1 knockdown would exert the opposite effect to putrescine supplementation, leading to accelerated senescence. HBEC^Tet−CDC6 cells were transfected with siRNA targeting ODC1 (siODC1) or a luciferase siRNA (siLuc) as a control, and exposed to DOX for 1 or 2 days to induce CDC6 expression (Tet ON), or left untreated (Tet OFF). Immunoblot analysis confirmed effective siRNA-mediated ODC1 knockdown in both Tet conditions (Fig. 6 A). Quantitative assessment of senescence on day 1 of CDC6 induction showed elevated SA-β-gal activity in siODC1 relative to siLuc-transfected cells which did not differ from Tet OFF controls (Fig. 6B). By day 2, siLuc-transfected Tet ON cells exhibited a nearly 2-fold increase in SA-β-gal activity (Fig. 6B), similar to parental cultures (see Fig. 4B). However, SA-β-gal activity increased by 2.7-fold in ODC1-depleted cells compared to Tet OFF controls (Fig. 6B). The effects of ODC1 knockdown on CDC6-induced senescence was mirrored by increased accumulation of TP53 relative to siLuc control cultures (Fig. 6 A). These findings establish ODC1, the rate-limiting enzyme for the intracellular ornithine to putrescine conversion, as a physiological inhibitor of CDC6-induced senescence.

Fig. 6.

Knockdown of ODC1 accelerates CDC6-induced senescence. (A) The knockdown of ODC1 exaggerates CDC6-induced TP53 accumulation. Panel Ai shows a validation of siRNA-mediated knockdown of ODC1 by immunoblot at various time intervals of doxycycline (DOX) treatment as well as TP53 levels. The mean ± SD fold changes of CDC6, TP53 and ODC1 expression levels from 3 independent experiments is shown in panel Aii. Statistical analysis was performed using one way-ANOVA and Tukey’s test. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. (B) Bar chart showing the fold changes in SA-β-Gal activity in HBECTet−CDC6 cells depleted of ODC1 following addition of DOX (Tet ON) for 1 or 2 days versus uninduced, Tet OFF cells. siLUC; siRNA against the unrelated luciferase gene. Statistical analysis was performed using one way-ANOVA and Sidak’s test. **p < 0.01 and ****p < 0.0001

The effects of CDC6 on MYC-ODC1 axis are mediated by ERK and GSK3β activation signals

The expression levels of MYC are affected co-translationally by several phosphorylation events. Thus, ERK-mediated phosphorylation of MYC at serine 62 (Ser⁶²) leads to MYC stabilization, whereas phosphorylation at threonine 58 (Thr⁵⁸) by GSK3β triggers MYC ubiquitination by the SCF(FBW7) ubiquitin ligase and proteasomal degradation [45].

As depicted in Figs. 5B and 7 A, MYC expression transiently increases at the early stages of CDC6 induction but falls below basal levels by day 3 and is nearly undetectable on day 6, indicative of proteasome-dependent degradation. Indeed, addition of the proteasomal inhibitor MG132 restored MYC expression levels with concomitant upregulation of ODC1 (Suppl. Figure 5). These observations prompted us to examine putative roles of ERK and GSK3β kinases in CDC6-mediated regulation of MYC expression. Immunoblot analysis of lysates isolated from Tet ON HBEC^Tet−CDC6 cultures revealed an early induction in ERK phosphorylation at day 1 of DOX treatment and a reduction thereafter. In contrast, GSK3β phosphorylation progressively increased only from day 2 onward (Fig. 7 A and Suppl. Figure 1D). Using antibodies specific for MYC phosphorylated at Ser⁶², we observed a transient Ser⁶² phosphorylation on day 1 sustained on day 2 and reduced below basal levels on days 3 and 6 of DOX treatment. In contrast, Thr⁵⁸ MYC phosphorylation levels were slightly elevated on day 2, increased further on day 3 and declined thereafter, in line with the progressive depletion of MYC expression (see Fig. 7A A). We also observed induction of p65 NF-κB phosphorylation by CDC6, albeit at a later time point than either ERK or GSK3β (Fig. 7A and Suppl. Figure 1D), suggesting that this pathway may exert roles apart from the MYC-ODC1 axis regulation.

Fig. 7.

CDC6 overexpression leads to upregulation of MYC and ODC1 via the MEK/ERK pathway. (A) DOX-induced overexpression of CDC6 in HBECTet−CDC6 cells activates an early ERK pathway response, typified by elevated levels of phosphorylated ERK1/2 (pERK1/2), followed by GSK3β and P65/RelA NF-κΒ phosphorylation (pP65 NF-κΒ). Activated ERK and GSK3β modulate phosphorylation of MYC at Ser62 (S62) and Thr58 (T58), respectively, that are known to affect MYC stability [44, 45, 57]. ACTB levels were assessed as loading control. (B) Treatment of HBECTet−CDC6 cells with 10µΜ of MEK inhibitor PD98059 for 24 h suppresses CDC6-mediated MYC and ODC1 levels. Data shown is representative of 3 independent experiments. (C) Supplementation of Tet-ON HBECTet−CDC6 cell cultures with putrescine that restores MYC expression (see Fig. 5 C), diminishes the effects of CDC6 on GSK3β and recovers activation of ERK. (D) Proposed model of the effects of putrescine on replication stress-induced cell fate decisions. Activation of the replication licensing factor CDC6 leads to an initial proliferative response followed by progressive accumulation of TP53, which is required for CDC6-mediated entry to senescence [11]. The early, proliferative phase of CDC6 induction is associated with elevated levels of MYC, driven by ERK-dependent phosphorylation at Ser62, and MYC-mediated ODC1 transactivation. The ensuing increase in putrescine biosynthesis protects HBECs from TP53-dependent induction of senescence. However, prolonged overexpression of CDC6 disrupts the MYC-ODC1-polyamine cycle because of GSK3β-mediated phosphorylation of MYC at Thr58 which targets it for proteasomal degradation. As MYC levels decline, ODC1 expression and intracellular putrescine are reduced, enabling TP53-dependent entry to senescence. Dotted arrows denote weak effects, bold arrows denote strong effects

To determine causative links between ERK activation signals and MYC upregulation occurring at the early stages of CDC6 induction, we treated HBEC^Tet−CDC6 cultures with 10µM of the MEK inhibitor PD98059 or with control vehicle (DMSO) and exposed them to DOX for 1 day in all possible combinations. Immunoblot analysis confirmed the suppression of CDC6-mediated activation of the MEK-ERK pathway and of MYC and ODC1 upregulation upon PD98059 treatment (Fig. 7B and Suppl. Figure 1E).

Given the impact of putrescine on the reversal of the CDC6-mediated MYC depletion (Fig. 5C and Suppl. Figure 1 F), we analysed ERK and GSK3β phosphorylation in Tet ON and Tet OFF cells exposed to varying concentrations of putrescine. As shown in Fig. 7C, supplementation of CDC6-overexpressing HBEC cultures with putrescine increased p-ERK and decreased p-GSK3β in a concentration-dependent manner, without affecting CDC6 expression. We conclude that CDC6 mediates ERK and GSK3β activation signals to regulate the MYC-ODC1 axis.

Single-cell transcriptomic data analysis recapitulates hallmarks of CDC6-induced senescence

Previous studies have reported increased epithelial cell senescence in the lungs of patients with severe COVID [46–48]. To assess the in vivo relevance of the CDC6–MYC–ODC1 axis indentified in our in vitro model, we re-analyzed publicly available single-cell RNA-sequencing (scRNA-seq) datasets derived from primary human alveolar epithelial cells obtained via lung resections or bronchoalveolar lavage fluid of healthy donors and patients with pneumonia of viral (severe COVID-19), viral/fungal (COVID-associated pulmonary aspergillosis; CAPA), or bacterial (non-COVID) etiology [31–34] and constructed a focused single-cell atlas of alveolar epithelial cells across these clinical groups (Fig. 8 A).

Fig. 8.

Representation of MYC-ODC1 metabolic rewiring using single-cell (sc) transcriptomic analysis of human alveolar epithelial cells. (A) Uniform manifold approximation and projection (UMAP) representation of human alveolar epithelial cells derived from healthy donors (HD; n = 3), or patients with pneumonia upon infectious insult: (i) Severe COVID, n = 20, (ii) COVID-associated pulmonary aspergillosis; CAPA, n = 14 or (iii) non-COVID pneumonia; NCP, n = 2. (B) Violin plots of senescence-associated secretory phenotype (SASP) marker genes expression across all groups. Projection of CDC6 (C1), MYC (D1) and ODC1 (E1) expression across scRNA-seq atlas and corresponding sample-specific signatures (C2, D2, and E2, respectively). **p < 0.01, ***p < 0.001

We observed enrichment of SASP gene signatures in alveolar epithelial cells from severe COVID-19 patients in line with previous reports [31–34] and extended it to CAPA and bacterial pneumonia (Fig. 8B). Strikingly, within SASP-positive epithelial cells, we detected a consistent increase in CDC6 mRNA expression in all pneumonia groups, including COVID-19 and CAPA (Fig. 8 C). To our knowledge, this is the first demonstration of CDC6 upregulation in senescence-like epithelial cells in human lung tissue during acute infection, suggesting that replication licensing stress may be a feature of pulmonary epithelial responses to injury or inflammation. Notably, these CDC6-expressing, SASP-positive cells also exhibited reduced expression of MYC and ODC1, recapitulating the molecular profile of CDC6-induced senescence observed in vitro (Fig. 8D and E).

Discussion

Building on previous work identifying CDC6 overexpression as a driver of replication stress and senescence in non-transformed cells [7–10], we have herein delineated metabolic contributors to replication stress-induced senescence. We observed a biphasic response of human bronchial epithelial cells to CDC6 overexpression, entailing an initial proliferative effect, followed by progressive cell cycle arrest and accumulation of the senescence marker SA-β-gal (Fig. 1). Through targeted metabolomic profiling we identified changes in both intracellular and extracellular metabolites, among which the polyamine putrescine emerged as a potential regulator of the senescence response to CDC6 overexpression (Fig. 2).

The polyamine pathway has been previously implicated in aging and age-related diseases. Rare genetic variants affecting enzymes in polyamine biosynthesis lead to metabolic imbalances collectively known as polyaminopathies, including Snyder–Robinson syndrome, Deoxyhypusine Synthase Disorder and Bachmann–Bupp syndrome [49]. Among polyamines, spermidine, generated from putrescine or via oxidative degradation of spermine, naturally declines with age and this reduction has been linked to the progression of neurodegenerative diseases, cardiovascular disorders, and metabolic dysfunctions [50, 51]. Conversely, spermidine supplementation has been shown to promote longevity and improve healthspan in various model organisms by enhancing autophagy, mitochondrial function, and stress resistance [52, 53]. Additional benefits of spermidine include inhibition of pro-inflammatory cytokine synthesis [23, 54], which is relevant to SASP and inflammaging [5, 55]. Epidemiological studies further support these findings, associating higher dietary spermidine intake with lower all-cause mortality in humans [56].

However, the role of putrescine in mitigating senescence in mammalian cells has remained largely undefined. Data presented herein show that CDC6 overexpression induces an early, transient increase in intracellular putrescine levels followed by a progressive decline below basal levels coincident with senescence entry (Figs. 2A and 5A). Functionally, putrescine supplementation attenuates CDC6-induced senescence (Fig. 4) whereas the knockdown of ODC1, the rate-limiting enzyme for the conversion of ornithine to putrescine [41], accelerates it (Fig. 6). These findings uncover a previously unrecognized role for putrescine as a modulator of the senescence response, extending the significance of polyamine metabolism to the regulation of cellular fates under genotoxic stress.

Mechanistically, our data support a model in which the MYC–ODC1–putrescine axis serves as a temporal regulator of the transition from proliferation to senescence. During the initial proliferative phase triggered by CDC6 induction, MYC is transiently upregulated, coincident with ERK-mediated MYC phosphorylation at Ser⁶² (Fig. 7 A & B) [44, 57]. Concordantly, ODC1 protein and intracellular putrescine increase at these early time points (Fig. 5 A), consistent with the established role of MYC as a transcriptional activator of ODC1 [42, 43]. However, as replication stress persists, GSK3β becomes activated and phosphorylates MYC at Thr⁵⁸ (Fig. 7 A), targeting it for proteasomal degradation [44, 45, 57]. As a result, both ODC1 expression and intracellular putrescine decline (Fig. 5A–B), creating a metabolic environment permissive for TP53 stabilization and activation of the senescence program (Fig. 7D) [11]. This is supported by the observation that exogenous putrescine attenuates both TP53 accumulation and SA-β-gal activity whereas ODC1 knockdown exerts the opposite effects (Fig. 4 C), implicating polyamine depletion as a regulatory node in replication stress-induced senescence.

Our findings align with and extend recent work by Afifi et al., demonstrating that reduction in MYC is necessary for transition to senescence following dual inhibition of MEK and CDK4/6 [44]. This study further showed that the duration, rather than the intensity, of oncogenic perturbation determines the probability of commitment to senescence, suggesting that it is not simply the maximal level of CDC6 expression that orchestrates senescence but the sustained duration of its induction which progressively destabilizes MYC and dismantles pro-proliferative and anti-senescent programs. Complementing this, Venkatesan et al. [58], using the same HBEC^Tet−CDC6 model, demonstrated that CDC6-driven replication stress activates a Chk1-dependent APOBEC3 program that promotes chromosomal instability in parallel with senescence. Together, these observations indicate that replication stress may elicit mechanistically separable outputs in epithelial cells: a mutagenic arm associated with chromosomal instability [58, 59] and a metabolic/tumor-suppressive arm marked by progressive loss of MYC and depletion of ODC1/putrescine (this study).

Whether and how malignant cells adapt to circumvent this metabolic barrier remains an important question. Notably, ODC1 overexpression has been causally linked to heightened cell proliferation [60] but only in tumor cells expressing high levels of MYC [61]. At the other end of the spectrum, we observed that CDC6-overexpressing HBECs that have escaped senescence and acquired features of epithelial to mesenchymal transition [11], exhibit persistently low levels of MYC, ODC1, and intracellular putrescine (Suppl. Figure 8). Interestingly, these cells no longer rely on CDC6 expression for their proliferative capacity and upregulate oncogenic programs including MDM2 overexpression to suppress TP53-mediated checkpoints [59]. Together, these observations highlight a remarkable metabolic and transcriptional plasticity that enables senescence evasion.

Findings reported herein may have implications beyond oncogenesis. We detected senescence-associated gene expression and increased CDC6 levels in alveolar epithelial cells from patients with severe COVID-19 pneumonia and CAPA, suggesting a stress-induced transcriptional program reminiscent of in vitro CDC6-induced senescence (Fig. 8). These cells also show downregulation of MYC and ODC1 alongside SASP gene enrichment, pointing toward a conserved CDC6–MYC–ODC1 regulatory axis in epithelial stress responses. In the case of CAPA, senescent epithelial cells characterized by impaired barrier integrity, altered innate defense, and chronic inflammation, may create a permissive niche for fungal colonization or tissue injury. While our transcriptomic data cannot establish causality or identify direct inducers of senescence in vivo, they reveal a distinctive molecular signature in severe pneumonia that warrants further mechanistic exploration. Future studies should investigate whether replication stress-induced senescence actively contributes to lung pathology or simply reflects the extent of epithelial damage in infectious contexts.

The mechanism(s) by which CDC6 overexpression triggers ERK and GSK3β activation signals is subject to ongoing investigations. One possibility is that the double strand breaks caused by CDC6-mediated replication fork collapse [11, 12] (also manifested by elevated γΗ2Ax; Fig. 5Bi) may activate RIP kinase 1 (RIPK1). Previous studies have demonstrated that upon DNA damage, including responses to proliferation-associated genomic instability [62], RIPK1 mediates ATM-dependent but TP53-independent autocrine [63] or cell-intrinsic signals [64] activating ERK and Akt-GSK3β [65, 66].

In addition to putrescine, sustained overexpression of CDC6 elicits significant reductions in intracellular levels of the aminoacids proline, valine and leucine (Fig. 2A), which have also been implicated in regulation of senescence. Thus, proline supplementation restores mitochondrial function and reverses several hallmarks of aging in senescent mesenchymal stem cells, including SA-β-Gal activity and SASP [67]. In yeast, elevated intracellular proline levels are linked to extended replicative lifespan, whereas proline depletion exerts opposite effects [68], suggesting a conserved role in aging and senescence. Declines in branched-chain amino acids (BCAAs) such as valine and leucine have been observed during replicative senescence in endothelial cells [69] and correlated with aging in Drosophila [70]. Interestingly, several enzymes involved in proline biosynthesis, including PYCR1 and ALDH18A1, and BCAA transporter genes such as SLC7A5 and SLC43A1 are direct transcriptional targets of MYC [71]. Therefore, MYC depletion ensuing from prolonged CDC6 overexpression may play a broader role in replication stress-induced senescence by coordinating the suppression of both polyamine biosynthesis and amino acid metabolism and uptake. Their relative contributions and potential synergies to CDC6-induced senescence in HBECs are subject to ongoing investigations.

The metabolomic profiling of CDC6-overexpressing senescent HBECs also revealed elevated glucose and reduced lactate levels in the culture supernatants (Fig. 2B), indicating impaired glucose uptake and glycolytic flux. This metabolic phenotype aligns with the dramatic reduction in the expression of MYC (Fig. 5B), a key transcriptional regulator of glucose metabolism-related genes [72–76]. Thus, the progressive loss of MYC expression in CDC6-overexpressing HBECs likely contributes to the suppression of glycolytic enzymes and glucose transporters, leading to decreased glucose utilization.

Interestingly, modulation of glucose availability did not significantly alter the kinetics or magnitude of CDC6-induced senescence (Fig. 3). This contrasts with reports in fibroblasts, where glucose restriction delays senescence and extends replicative lifespan [38]. However, senescence-associated metabolic remodeling of glycolysis is highly context-dependent [17]. For instance, senescence induced by p16^Ink4a in pancreatic β-cells enhances glucose-stimulated insulin secretion but is not driven by glucose metabolism per se [77]. Taken together, these findings suggest that glucose metabolism may be dispensable for senescence induction in certain epithelial and endocrine cell types and/or senescence-inducing stimuli.

Collectively, our work uncovers a previously unrecognized role for the polyamine pathway in modulating replication stress-induced senescence and suggests a broader metabolic network underpinning cellular fate decisions in response to oncogenic stimuli. Our data also highlight MYC as a temporal rheostat linking replication stress to metabolic responses that influence the transition from cell proliferation to senescence. Targeting the polyamine biosynthesis pathway may thus represent a novel strategy to modulate senescence and mitigate age-related pathologies.

Author contributions

Theodora Vasilogiannakopoulou performed experiments and wrote a draft manuscript. Olga Begou performed the metabolomic analysis and wrote part of the manuscript. Christina Simoglou Karali performed the single cell transcriptomic data analysis and revised the manuscript. Georgia Efthymiou supported the experiments. Maria Roubelakis, Vassilis Gorgoulis and Kalliopi K. Gkouskou revised the manuscript. Helen Gika supervised the metabolomic analyses and revised the manuscript. Aristides G. Eliopoulos designed and supervised the study and finalized the preparation of the manuscript.

Funding

This work was supported by the Hellenic Foundation for Research and Innovation (H.F.R.I.) grant number 3782 and acronym “PaCoRel” to A.G.E and V.G, and the action “Flagship Research Projects in challenging interdisciplinary sectors with practical applications in Greek Industry”, implemented through the National Recovery and Resilience Plan Greece 2.0 and funded by the European Union – NextGenerationEU, project code: TAEDR-0541976, and acronym “PRO-sCAP” to A.G.E.

Data availability

Any data reported in this paper is available from the corresponding author upon reasonable request.

Declarations

Ethical approval

None to declare.

Consent for publication

All authors consented to publish the manuscript.

Competing interests

A.G. Eliopoulos and K. Gkouskou are co-founders of GENOSOPHY S.A., a genomics spin-off company of the National and Kapodistrian University of Athens. This does not affect the research results of this paper.