Mitochondrial bioenergetics-SASP crosstalk determines senolytic efficacy in therapy-induced senescence
Àngela Llop-Hernández, Sara Verdura, Júlia López, Begoña Martin-Castillo, Javier A. Menendez, Elisabet Cuyàs
TL;DR
This study shows how mitochondrial energy production and inflammation interact to determine how well senolytic drugs work in cancer cells undergoing therapy-induced senescence.
Contribution
The study identifies a mitochondrial-inflammation crosstalk that governs senolytic drug efficacy in senescent cancer cells.
Findings
Mitochondrial bioenergetic flexibility influences senolytic permissiveness in cancer cells.
Baseline succinate oxidation predicts inherited senolytic thresholds in TIS cells.
Inflachromene can decouple mitochondrial function from senolytic response, creating drug-resistant senescent cells.
Abstract
Mitochondria integrate senescence and apoptotic fates, yet it is unclear whether their ability to oxidize different fuels for energy production influences their vulnerability to senolytics in therapy-induced senescence (TIS). Using MitoPlates™ technology, we functionally mapped the mitophenotypes of TIS cancer cells by quantifying electron transport chain (ETC) flux from various NADH/FADH2 substrates. We then related these profiles to the responsiveness of TIS cancer cells to BCL-xL-targeting BH3 senolytics, as well as to inflammatory SASP signaling sensed by an NF-κB/miR-146a reporter. Mechanistically distinct senogenic stressors produced markedly different bioenergetic outputs and substrate diversity, establishing mitochondria as an emergent, stress-encoded property of TIS phenomena. Increased mitochondrial bioenergetic flexibility corresponded with senolytic permissiveness within…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7- —Ministerio de Ciencia e Innovación/Spanish Research Agency (MCIN/AEI): Grant PID2022-141955OB-I00
- —https://doi.org/10.13039/501100004587Ministry of Economy and Competitiveness | Instituto de Salud Carlos III (Institute of Health Carlos III)
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsTelomeres, Telomerase, and Senescence · Acute Myeloid Leukemia Research · Retinoids in leukemia and cellular processes
Introduction
Alterations in numerous aspects of mitochondrial biology, including molecular features, activity, function, and dynamic behavior, are well-established features of cellular senescence [1]. Specific mitochondrial traits play a driving role in establishing senescence-like growth arrest and, particularly, in developing the senescence-associated apoptosis resistance and senescence-associated secretory phenotype (SASP) [2–7]. An increase in mitochondrial mass, mitochondrial DNA, and mitochondrial production of reactive oxygen species (ROS), as well as a decrease in mitochondrial membrane potential, accompany the onset of senescence [8]. Indeed, manipulations that lead to impaired mitochondrial functions promote cellular senescence with a modified SASP and a lower NAD^+^/NADH ratio [9]. Conversely, mitochondrial depletion prevents the development of senescence features, including the expression of the cyclin-dependent kinase (CDK) inhibitor p16^INK4a^, cell enlargement, SA-β-galactosidase (SA-β-gal) activity, and the SASP [5–7, 10]. Alterations in mitochondrial energy metabolism also appear to play a critical role in inducing senescence. Retinoblastoma protein (RB)-mediated glycolytic activity activates a metabolic flux toward mitochondrial oxidative phosphorylation to increase the levels of TCA cycle intermediates in senescent cells [11]. Increased fatty acid oxidation (FAO) results in high basal respiratory activity rates in cells that have undergone senescence [12]. Blocking carnitine palmitoyltransferase 1, the rate-limiting step in mitochondrial fatty acid oxidation, restores a pre-senescent metabolic rate and selectively blocks the inflammatory SASP that accompanies senescence [12].
Most, if not all, of the observed links between alterations in mitochondrial bioenergetics and senescence are based on studies of oncogene-induced and replicative senescence. However, the characteristics and significance of mitochondrial bioenergetic functions in therapy-induced senescence (TIS) are not well understood. TIS is a form of accelerated cellular senescence induced by oxidative, replicative, mitotic, or genotoxic anticancer therapeutics that affects both malignant and normal cells [13–17]. TIS can be intentionally utilized as a tumor suppressive mechanism that limits cellular proliferation and enhances anti-tumor immunity [18–22]. However, the persistent accumulation of TIS cancer cells may paradoxically promote tumorigenesis by promoting bystander cell proliferation, treatment resistance, and metastatic progression through mechanisms such as apoptosis escape and cellular dormancy [23–26]. TIS in otherwise healthy tissues is also an off-target effect of cancer therapy that exacerbates toxicities such as myelosuppression, organ fibrosis, inflammation, frailty and fatigue. This contributes to clinically significant, treatment-related, iatrogenic sequelae [27–30]. TIS may render cancer cells susceptible to synthetic lethal approaches that target permanently or transiently acquired senescence-associated vulnerabilities. As the senescent state was thought to be a universally primed apoptotic state, it is not surprising that pro-apoptotic BH3 mimetics are the most widely used category of TIS-targeting senolytic agents to date [31–40]. These agents were originally designed to selectively antagonize anti-apoptotic BCL-2 family proteins and enhance the functional stoichiometry of pro-apoptotic BH3-only proteins. Nevertheless, some TIS cancer cells can exhibit primary refractoriness to ABT-263/navitoclax, the most extensively studied dual BCL-xL/BCL-2 inhibitor with broad-spectrum senolytic activity [41–47].
Recently, our group and others have shown that, contrary to the prevailing dogma, the mitochondria of TIS cancer cells are less primed for apoptosis than those of proliferating precursors. Despite this, they share a conserved, druggable dependence on specific members of the BCL-2 family, particularly BCL-xL, for survival [48–52]. Interestingly, the pre-existing priming and anti-apoptotic dependencies of parental mitochondria in non-senescent cells are retained upon induction of senescence. In this study, we investigated further whether the overall bioenergetic capacity and/or the ability of mitochondria to flexibly switch between alternative bioenergetic fuels affects the sensitivity of TIS cancer cells to senolytics. Furthermore, we hypothesized that senolytic-resistant TIS cancer cells could exhibit a distinct, mitochondria-regulated SASP profile. To address this question, we used the so-called MitoPlates S-1™, a cell-based phenotyping platform that measures electron flow rates into and through the electron transport chain (ETC) from NADH- and FADH_2_-producing mitochondrial metabolic substrates [53–57]. We combined this cell-level functional phenotyping of mitochondrial metabolic flexibility with SASP profiling in the extracellular environment of TIS cancer cells that were engineered to express an inflammation-sensing miR-146a promoter construct [58, 59].
Our results suggest that TIS mitochondria switch between alternative fuels for energy production differently in response to mechanistically diverse senescent triggers. However, the apparent correlation between heightened mitochondrial flexibility and the likelihood of sensitivity to BCL-xL-targeting BH3 mimetics depends largely on the bioenergetic configuration of the mitochondria in pre-senescent, parental cells. This configuration acts as a metabolic fingerprint that restricts the magnitude of senolytic responsiveness. Furthermore, we demonstrate that directly preventing the establishment of an inflammatory SASP by inhibiting the chromatin remodelers HMGB1/2 with the small compound inflachromene [60–63] dominates over mitochondrial bioenergetics in driving senolysis resistance in senescent cancer cells. The senolytic response of TIS cancer cells appears to be regulated by a multi-layered circuit in which mitochondrial bioenergetic legacy shapes qualitative senolytic permissiveness, bioenergetic flexibility gained during TIS sets an upper quantitative senolytic limit, and mitochondria-inflammatory SASP cross-talk seems necessary for senolysis. Monitoring the interaction between mitochondrial metabolic (in)flexibility and the inflammatory output of the SASP could be used to predict and improve senolytic responses in TIS cancer cells.
Results
The global bioenergetic capacity of TIS mitochondria varies depending on the nature of the mechanistic TIS stressor
Senescence is associated with dramatic changes in the mitochondrial phenotype, including mitochondrial mass, dynamics, structure, function, and apoptotic priming [2, 4–6, 9, 48–52, 64]. However, it is unclear whether changes in the overall bioenergetic activity of TIS mitochondria at the cellular level are a consistent characteristic of the TIS phenomenon or if they vary in response to distinct senescence-inducing stressors.
We initially chose the A549 type II-like alveolar lung adenocarcinoma cell line to screen mitochondrial function across TIS phenotypes because it has become the reference cell model for rapid and reproducible TIS protocols [41, 51, 65–67]. Four mechanistically distinct TIS inducers were used: bleomycin (an antibiotic that promotes oxidative stress), alisertib (an Aurora A kinase inhibitor that promotes mitotic stress [31, 41, 68]), doxorubicin (a topoisomerase II inhibitor and DNA intercalator that induces DNA damage stress [46, 69, 70]), and palbociclib (an inhibitor of cyclin-dependent kinases 4 and 6 [CDK4/6] that induces replication stress [71–76]). After seven days of treatment, the A549 cells likewise acquired hallmark features of senescence, including a marked cytomorphological remodeling (e.g., enlarged and flattened morphology), SA-β-gal-positive phenotypes (Fig. 1A), and cell cycle arrest-related loss of Ser-807/811 retinoblastoma (RB) phosphorylation, downregulation of proliferating cell nuclear antigen (PCNA), and accumulation of p21^WAF1/Cip1^ (Fig. S1A).Fig. 1. Phenotyping of mitochondrial function in TIS A549 cancer cells.A Senescent phenotype of TIS A549 cancer cells. Representative images (n = 3) of SA-β-gal staining in A549 lung cancer cells were taken after treatment with senescence-inducing concentrations of palbociclib (5 μmol/L), doxorubicin (50 nmol/L), alisertib (500 nmol/L) or bleomycin (20 μmol/L) for 7 days (scale bar: 500 μm for proliferative/untreated cells; 200 μm for TIS cells). B Top. The overall capacity and the type of metabolic substrates used by mitochondria to generate NADH and FADH_2_ were evaluated using mitochondrial phenotyping arrays (Mitoplate™ S-1), which consist of 96-well plates precoated with low millimolar concentrations (2–5 mmol/L) of specific mitochondrial substrates. When permeabilized proliferating and TIS counterparts are added to the plate without any substrate other than that present in each well, the redox dye reduction rate provides a measure of the rates of electron flow into and through the ETC from cytoplasmic, TCA cycle, and other metabolic substrates that generate NADH or FADH_2_. Electrons donated to complex I or complex II travel to the distal end of the ETC where a tetrazolium redox dye acts as a terminal electron acceptor and changes from colorless to a purple formazan upon reduction. Assay plates contain individual 31 lyophilized cytoplasmic and mitochondrial metabolic substrates, and wells are repeated in triplicate. Cytoplasmic substrates include glucose, glycogen, glucose-1-P, glucose-6-P, gluconate-6-P, glycerol-P, and lactic acid; TCA cycle substrates include pyruvic acid, citric acid, isocitric acid, aconitic acid, α-ketoglutaric acid, β-hydroxybutyric acid, glutamic acid, glutamine, alanine-glutamine, serine, ornithine, tryptamine, and malic acid; and other mitochondrial substrates include acetyl-carnitine + malic acid, octanoyl-carnitine + malic acid, palmitoyl-carnitine + malic acid, pyruvic acid + malic acid, amino-butyric acid + malic acid, ketoisocaproic acid + malic acid, leucine + malic acid. Bottom. Figures show kinetic heat maps of the metabolic substrate consumption (up to 6 h) in proliferative and palbociclib, doxorubicin, alisertib, and bleomycin TIS cancer cells. Values shown are the mean optical density at 590 nm of n ≥ 3 independent experiments performed in triplicate. Sum of the signals (AUCs) of all mitochondrial substrates/intermediates representing the total metabolic activity of each TIS phenotype, integrated over time, is shown (global mitochondrial metabolic score). Graphs represent the means (columns) ± SEM (bars) of ≥3 independent experiments. Statistically significant differences (ANOVA analysis) are shown.
We characterized these four A549 TIS phenotypes using the Biolog MitoPlate™ S-1, a phenotypic metabolic array measuring electron flow rates through the electron transport chain (ETC) of various metabolic substrates producing either NADH (e.g., pyruvate, L-malate, α-ketoglutarate, D-isocitrate, L-glutamate, D-β-hydroxybutyrate) or FADH_2_ (e.g., succinate, α-glycerol-3-P) [53–57]. Table S1 summarizes the transporter and dehydrogenase genes and proteins involved in the processing of the NADH- and FADH_2_-producing mitochondrial substrates. The electrons donated to complexes I or II travel to the distal end of the ETC, where a tetrazolium dye acts as a terminal electron acceptor. The dye changes from colorless to a purple formazan upon reduction (Fig. 1B, top). We delineated the mitochondrial functional phenotype by adding saponin-permeabilized TIS/parental proliferative cells and a redox dye to 96-well microplates precoated with different lyophilized mitochondrial substrates. Dye accumulation, which reflects the capacity for electron transport and respiratory capacity and serves as a proxy for mitochondrial respiratory capacity at the cellular level, significantly varied across TIS phenotypes. Kinetic analysis of scaled values of the bioenergetic utilization of 31 mitochondrial substrates producing NADH and FADH_2_ strongly suggested that certain A549 TIS derivatives have increased energy production compared to proliferative A549 parental cells, both globally and in specific pathways (Fig. 1B, bottom). Total bioenergetic capacity and mitochondrial substrate diversity were lowest in palbociclib-induced replicative TIS, intermediate in doxorubicin-induced genotoxic and alisertib-induced mitotic TIS, and highest in bleomycin-induced oxidative TIS. We confirmed that the sum of the signals of all mitochondrial substrates/intermediates—representing the total metabolic activity integrated over time and reflecting the global TIS mitochondrial activity or the absolute bioenergetic output of TIS mitochondria at the cellular level—depended on the type of TIS-inducing stress by calculating the AUC of each metabolite (Fig. 1B).
Together, these data show that the TIS phenomenon does not establish a consistent mitochondrial bioenergetic state at the cellular level. Rather, the ETC substrate utilization and global bioenergetic output are variable, mechanistically imprinted features of TIS stressors.
The mitochondrial flexibility of TIS phenotypes varies depending on the type of TIS stressor
We then organized the bioenergetic processing of mitochondrial metabolites by their association with specific metabolic functions (Fig. 2A). The bleomycin and alisertib TIS phenotypes exhibited greater global bioenergetic generation from a higher number of mitochondrial metabolites to generate energy-rich NADH and FADH_2_ than doxorubicin and palbociclib TIS phenotypes (Fig. 2B). Glycolytic intermediates such as D-glucose-1-phosphate and D-glucose-6-phosphate, as well as the pentose phosphate pathway (PPP) intermediate D-gluconate-6-phosphate, produce higher levels of NADH/FADH_2_ in alisertib- and bleomycin-induced TIS cancer cells than in doxorubicin- and palbociclib-induced TIS cancer cells. A significant difference in the utilization of substrates that directly activate or enter the mitochondrial TCA cycle and the ETC was also observed in alisertib- and bleomycin-induced TIS cancer cells compared to their proliferative counterparts. Interestingly, an increased bioenergetic processing of metabolites associated with the malate-aspartate shuttle (MAS), indirect TCA activation, amino acid catabolism, and, in particular, fatty acid β-oxidation (FAO), appeared to be a distinctive hallmark of the mitochondrial metabolic phenotype in the bleomycin TIS cells (Fig. 2B).Fig. 2. Pathway-specific utilization of mitochondrial bioenergetic metabolites in TIS A549 cancer cells.A Map of the NADH- and FADH_2_-producing bioenergetic pathways included in the Biolog Mitoplate™ S-1. Since these bioenergetic processes are directly linked to higher expression levels of transporters and/or enzymes involved in the utilization of specific substrates (see Table S1), this approach can functionally delineate how active (and therefore physiologically relevant) a mitochondrial pathway is for a given TIS phenotype. B Left. The figure shows the heat map of the AUC values for Biolog Mitoplate™ S-1-based measurements of the utilization of 31 metabolites over 1–6 h (see Fig. 2), organized according to their association with specific metabolic functions. Red indicates higher utilization. Shown are the mean AUCs of n≥3 independent experiments performed in triplicate. Right. Pooled AUCs of processed mitochondrial bioenergetic metabolites according to their association with specific metabolic functions (A) in proliferative and TIS cancer cells. Graphs represent the means (columns) ± SEM (bars) of ≥3 independent experiments. Statistically significant differences (ANOVA analysis) are shown. C Flower model Venn diagrams showing common and TIS phenotype-specific mitochondrial metabolic substrates with AUCs ≥3.
Using a 3.0-fold change in AUC as a cut-off (none of the mitochondrial metabolites reached such an AUC increase in the proliferative counterparts), we found that all of the A549 TIS phenotypes overutilized five mitochondrial bioenergetic sources (Fig. 2C), namely: D-glucose-1-phosphate, D-glucose-6-phosphate, citric acid, L-malic acid, and succinic acid. Bleomycin and alisertib TIS cells shared an overutilization of fumaric acid, α-keto-glutaric acid, and D-gluconate-6-phosphate. Alisertib TIS cells uniquely overutilized D, L-isocitric acid. The bioenergetic utilization of malic acid in combination with pyruvic, γ-amino-butyric (GABA), and α-ketoisocaproic acid, or in combination with the fatty acid transporters acetyl- and octanoyl-carnitine, was restricted to the bleomycin-induced TIS cancer cells (Fig. 2C).
We ranked the fold changes obtained from the pairwise comparison of the energy generation by each mitochondrial substrate/intermediate in each A549 TIS phenotype relative to the proliferative parental cells (Fig. S2A). Then, we generated flower model Venn diagrams to determine whether the mitochondrial utilization of bioenergetic substrates across the TIS phenotypes could distinguish between core and TIS phenotype-specific mitochondrial metabolic sources that are significantly overutilized by the four TIS phenotypes (cut-off ≥ 5.0-fold change; Fig. S2B). Citric acid, the first intermediate formed in the TCA cycle, and pyruvate, the end product of glycolysis, were the preferred mitochondrial substrates/intermediates employed by all of the TIS phenotypes compared to their proliferative counterparts. Tryptamine, a tryptophan derivative and catabolic substrate of mitochondrial monoamine oxidase (MAO) enzymes [77–79], was a preferred bioenergetic metabolite shared by palbociclib- and bleomycin-induced TIS cancer cells. Bleomycin TIS cancer cells preferred up to ten bioenergetic metabolites, most of which are involved in amino acid catabolism and β-oxidation, over their proliferative counterparts (Fig. S2B).
TIS mitochondria display a shared core of overutilized substrates, notably pyruvate and citrate-linked inputs, but they diverge markedly in metabolic breadth and pathway preference across stressors. Oxidative TIS retain higher flexibility, as demonstrated by amino acid catabolism and β-oxidation in bleomycin-imprinted mitochondrial rewiring.
The correlation between mitochondrial (in)flexibility and senolysis by BCL-xL-targeting BH3 mimetics is limited by the bioenergetic features of the cell-of-origin mitochondria
A549 senescent cells are by far the most sensitive TIS model to BH3 senolytics, as they exhibit the greatest decrease in the half-maximal inhibitory concentration (IC_50_) values of the BCL-2/BCL-xL-targeted ABT-263/navitoclax when compared to a wide panel of cancer cell lines [41, 49, 51]. Using AlamarBlue™-based cell viability assays, we confirmed our previous observation that the ABT-263/navitoclax senolytic index (SI) of A549 TIS cells (IC_50 proliferative_/IC_50 TIS_) varies in a gradient manner depending on the TIS stressor (Fig. S3). The SI was very low with palbociclib TIS, intermediate with doxorubicin and alisertib TIS, and very high with bleomycin TIS. This trend was also evident when ABT-263/navitoclax was substituted with the BCL-xL-specific inhibitor A1331852. A1331852 senolytic indexes varied from less than two in the case of palbociclib TIS to greater than 100 in the case of bleomycin TIS (Fig. S3).
The increased sensitivity of A549 TIS cells to pro-apoptotic BH3 mimetics is associated with their high level of intrinsic mitochondrial apoptotic priming and their strong dependence on specific anti-apoptotic BCL-2 family proteins, such as BCL-xL [49, 51]. These features are present in the parental, pre-senescent state and persist and amplify upon acquisition of the senescent phenotype [49, 51]. In contrast, MCF-7 cells, a widely used in vitro model for studying breast cancer biology, have minimal reliance on BH3 targets and remain poorly responsive to BH3 mimetics when acquiring a senescent phenotype [41]. At the opposite end of the spectrum are LoVo colorectal cancer cells, which are completely unresponsive to senolytic BH3 mimetics. Their BAX-inactivating mutational background and resulting lack of mitochondrial priming abolish any senolytic effect, regardless of senescent state [41, 48, 51]. Thus, the susceptibility of TIS cancer cells to BH3 senolysis seems to reflect the intrinsic apoptotic wiring of the mitochondria of the cell-of-origin, rather than merely the ability to become senescent [49, 52].
To assess whether stressor-related changes in the mitochondrial functional phenotype of TIS cancer cells are similarly influenced by the nature of the parental cell line, we applied phenotypic metabolic arrays to MCF-7 cells rendered senescent with bleomycin, alisertib, doxorubicin, and palbociclib. After seven days, the cells exhibited characteristic indicators of senescence, including enlarged, flattened appearance, SA-β-gal positive staining, diminished Ser-807/811 phosphorylation of the RB associated with cell-cycle arrest, and downregulation of PCNA coupled to increased levels of p21^WAF1/Cip1^ (Fig. S1B). Similar to A549 cells, MCF-7 TIS cells’ mitochondria underwent distinct modifications in their overall bioenergetic capacity and substrate diversity depending on the type of TIS stressor (Fig. 3A). The alisertib TIS phenotype apparently exhibited the most flexible bioenergetic generation due to a greater number of mitochondrial metabolites. When mitochondrial metabolites were organized by their metabolic function, the bleomycin TIS phenotype was again associated with increased catabolic processing of amino acids and β-oxidation of fatty acids (Fig. S4A). Although alisertib TIS MCF-7 cells were the most sensitive to the BCL-xL-targeting BH3 mimetics ABT-263/navitoclax and A1331852, their senolytic indexes ranged from 2 to 3 (Fig. S3). This is several orders of magnitude lower than the robust senolytic responses observed in A549 cells.Fig. 3. Phenotyping of mitochondrial function in TIS MCF-7 and LoVo cancer cells.A, B Senescent phenotype of TIS MCF-7 and LoVo cancer cells. Representative images (n = 3) of SA-β-gal staining in MCF-7 breast cancer cells and LoVo colon cancer cells were taken after treatment with senescence-inducing concentrations of palbociclib (5 μmol/L), doxorubicin (50 nmol/L), alisertib (500 nmol/L) or bleomycin (20 μmol/L) for 7 days (scale bar: 200 μm). The panel also shows heat maps of the AUC values for Biolog Mitoplate™ S-1-based measurements of the utilization of 31 metabolites by TIS MCF-7 and LoVo cancer cells over 1-6 h. Red indicates higher utilization. The mean AUCs are shown for n≥3 independent experiments performed in triplicate. C Left. Succinate oxidation capacity (AUC values) of A549, MCF-7, and LoVo proliferative/untreated cancer cells and their TIS derivatives. The graph represents the means ± SEM of ≥3 independent experiments. Right. Heat maps showing the utilization of 31 metabolites over 1–6 h by A549, MCF-7, and LoVo proliferative/untreated cancer cells using Biolog Mitoplate™ S-1. The mean AUCs are shown for n ≥ 3 independent experiments performed in triplicate.
We then examined the impact of extreme mitochondrial unpriming on the functional bioenergetic phenotype of TIS mitochondria using BAX-mutated LoVo cells (Fig. 3B). These cells are completely refractory to pro-apoptotic BH3 mimetics [41, 48, 51]. Using palbociclib and alisertib as TIS inducers, we observed a slightly increased utilization of upper glycolytic and PPP metabolites, α-ketobutyric acid, and, in the case of alisertib TIS LoVo cells, the glycerol-3-phosphate (G3P) shuttle. Organizing the data by metabolic function revealed that only palbociclib TIS exhibited a significantly increased β-oxidation (Fig. S4B).
Notably, unlike the A549 and MCF-7 cell models, which exhibited significantly increased utilization of succinic acid (the MitoPlate™ assay positive control) across all the TIS phenotypes, LoVo cells exhibited slightly decreased succinic acid utilization when they acquired the senescent phenotype. These data suggested that the directionality of succinate utilization during TIS is not a universal feature of the senescence program. Since succinate feeds electrons directly into Complex II (CII) and the CoQ pool, bypassing upstream metabolic variability, its differential oxidation provides a functional indication of the organization, functionality, and stability of the pre-existing mitochondrial ETC. Thus, it may reflect the intrinsic, lineage-specific mitochondrial architectures present in the parental cells. We compared the baseline succinate utilization of the parental A549, MCF-7, and LoVo cells and their TIS derivatives (Fig. 3C, left). Cells with a higher baseline succinate oxidation (A549, followed by MCF-7) increased this flux further upon senescence. However, LoVo cells, which have the weakest intrinsic CII capacity, could not sustain succinate-driven ETC activity during senescence. The baseline succinate utilization profiles and the global bioenergetic fingerprints at the pre-senescent baseline state (A549» > MCF-7>LoVo; Fig. 3C, right) both positively correlated with the senolytic sensitivity of the TIS derivatives.
Together, these results suggest that stressor-specific mitochondrial bioenergetic rewiring occurs across different lineages during TIS, but its magnitude and direction are limited by the pre-existing mitochondrial bioenergetic program. The baseline ETC capacity of parental mitochondria, exemplified by succinate/CII oxidation, correlates with BCL-xL BH3 senolytic responsiveness, whereas mitochondrial (in)flexibility alone does not. Therefore, the inherited mitochondrial architecture appears to set the ceiling for TIS bioenergetics and senolytic vulnerability.
Mitochondrial bioenergetic flexibility is an emergent property of the TIS phenotype
We evaluated the mitochondrial substrate oxidation profiles of TIS cancer cells by examining the entire mitochondrial system within each cell using a high-dimensional substrate utilization fingerprint [53–57, 80]. This agnostic approach integrates mitochondrial mass, network architecture, organelle interactions, and substrate transport into a single functional phenotype per cell rather than by mitochondrial unit. To better understand how the entire mitochondrial system partitions flux between substrates in TIS cancer cells, we first confirmed that a 3- to 5-fold increase in the mitochondrial mass per cell in different A549 TIS phenotypes compared to their parental counterparts using the MitoView™ Green dye, a bright, photostable, and non-toxic fluorogenic mitochondrial stain for live cells. In the case of MCF-7 cells, all the TIS phenotypes uniformly increased their mitochondrial mass by approximately 2-fold (Fig. 4A).Fig. 4. Mitochondrial content and mitochondrial-related transcriptional adaptations in TIS A549 and MCF-7 cancer cells.A Left. Representative images (n = 3) of MitoView™ green mitochondrial staining in live proliferative and TIS cancer cells (scale bar: 500 μm). Right. Fold-change in mitochondrial content (FMC) was measured by flow cytometry as described in the “Materials and methods” section. B Transcript abundance of 13 mitochondria-related genes was calculated using the ΔΔCt method and presented as fold-change. Statistically significant differences between proliferative and TIS phenotypes are shown (*p < 0.05; **p < 0.005) using ANOV Aanalysis.
To further explore whether changes in pathway-specific bioenergetics across TIS phenotypes could be explained by transcriptional mitochondrial adaptations, we performed quantitative RT-qPCR on a curated panel of mitochondria-related genes (n = 13) involved not only in apoptosis (BCL2 and BCL2L1) but also in respiratory complexes, ATP synthase, TCA entry, and fatty acid import/β-oxidation (ACLY, CPT2, CPT1B, CS, ETFDH, MPC1, NDUFAB1, SDHA, SHMT2, SLC25 A1, and SLC25A20) (Fig. 4B). A549 TIS cells exhibited a similar decrease in BCL2 expression, accompanied by a significant BCL2L1 (BCL-xL/xS) upregulation across most of the TIS phenotypes. MCF-7 cells also showed notable BCL2L1 upregulation across all phenotypes, particularly in bleomycin TIS cells. TIS phenomena in A549 and MCF-7 cells provoked only discrete yet functionally coherent transcriptional adaptations in mitochondrial metabolism relative to proliferative controls. Across all senescence inducers, the vast majority of transcripts remained close to baseline levels of proliferative controls, confirming a global preservation of the transcriptional homeostasis of nuclear-encoded mitochondrial genes across mechanistic senescence modalities. However, a small subset of genes consistently crossed the significance threshold, indicating a highly selective, low-amplitude remodeling of the mitochondria-related transcriptome. In A549 cells, CPT2 and SLC25A20 were upregulated across all phenotypes but more markedly in bleomycin TIS cells. CPT2 encodes the mitochondrial inner membrane carnitine palmitoyltransferase II, which converts long-chain acyl-CoAs to acylcarnitines to allow them to enter the mitochondria. SLC25A20 encodes the carnitine-acylcarnitine translocase/carnitine acetyltransferase (CACT/CrAT), which shuttles those acylcarnitines into the mitochondrial matrix in exchange for free carnitine. This suggests a reinforcement of the carnitine shuttle and β-oxidation flux, which is compatible with the functional bioenergetic phenotype delineated with MitoPlates™. In MCF-7 cells, SLC25A20 upregulation was maintained, but CPT2 was substituted by CPT1B. CPT1B encodes the mitochondrial outer membrane carnitine palmitoyltransferase I, which catalyzes the first step in long-chain fatty acid import into mitochondria and is the rate-limiting step for β-oxidation of fatty acids. Once again, this coupled enhancement of CPT1B/SLC25A20 gene expression was notable across all phenotypes, especially in bleomycin TIS cells. MPC, which encodes for the mitochondrial pyruvate carrier, was similarly upregulated across all TIS phenotypes in A549 cells but not in MCF-7 cells (Fig. 4B).
These data support the idea that mitochondrial bioenergetic flexibility is an emergent, system-level feature of TIS. This flexibility arises from increased mitochondrial content and coordinated flux partitioning, rather than from uniform changes in single organelles. Despite the fact that nuclear-encoded mitochondrial transcript homeostasis is largely preserved, a small, coherent gene module centered on the carnitine shuttle/β-oxidation is selectively reinforced. This matches the functional substrate-use signatures across stressors.
miRNA-146a-sensed inflammatory SASP distinguishes senolytic responses across TIS phenotypes
Growing evidence suggests that mitochondrial metabolic remodeling plays a conductive role in SASP regulation by linking acetyl-CoA production to histone acetylation at SASP gene loci, promoting their expression [81–83]. However, although apoptotic stress itself has been identified as a regulator of the SASP [7, 84], it remains unclear whether the expression of SASP factors can predict the sensitivity of TIS cancer cells to senolytics. Among the classical SASP factors IL-6 (IL6) and IL-8 (CXCL8), which have been described as classical SASP factors that are consistently upregulated in normal cell senescence and play a role in maintaining and propagating SASP expression [85–87], only high CXCL8 expression has been found to predict senolytic efficacy [48]. Recently, the SASP of senolytic-resistant senescent cells has been shown to entail reduced production of pro-inflammatory/apoptotic factors, increased production of growth/fibrotic factors, and reduced inflammation in non-senescent cells [88].
We aimed to explore the link between the mitochondrial control of SASP phenotypes and the senolytic response. To achieve this, we used A549 and MCF-7 cells that were engineered to express the microRNA 146a (miR146a) promoter fused to eGFP [58, 59]. miR146a acts as a fine-tuning mechanism that prevents the overstimulation of the inflammatory response, thereby limiting the harmful effects of the high levels of SASP factors on surrounding tissues. This negative feedback loop primarily occurs through binding of the inflammatory, SASP-driving NF-κB transcription factor—the master regulator of inflammation and stress responses [58, 89–94]—to its binding sites at the miR146a promoter. First, to investigate whether the qualitative/quantitative composition of SASPs generated differed depending on the mechanistic nature of different TIS stressors, we performed cytokine arrays with conditioned media from proliferative and TIS A549 and MCF-7 cells using the Luminex® platform (Fig. 5A). In agreement with previous studies, all the TIS stressors generated SASP landscapes that were largely dictated by the baseline secretome of the cell-of-origin (Fig. S5) [41]. Interestingly, despite the presence of a qualitatively and quantitatively similar SASPs across all the TIS inducers, flow cytometry analysis of miR146a-eGFP+ cells revealed significant differences in the ability of the different TIS agents to activate the miR146a promoter in the corresponding TIS phenotypes (Fig. 5B). For doxorubicin, alisertib, and bleomycin TIS, a group of miR146a-positive (NF-κB-driven) phenotypes could be clearly distinguished from the miR146a-negative (non-NF-κB-driven) palbociclib phenotype in both A549 and MCF-7 cells. Finally, we explored the hypothesis that, if mitochondrial metabolism drives a miR146a promoter-(NF-κB)-sensed component of the SASP, then blocking acetyl-CoA production via mitochondrial β-oxidation [82, 95] would result in a loss of the miR146a-positive phenotype. Accordingly, treating bleomycin TIS A549 cells with increasing concentrations of the CPT1 inhibitor etomoxir, which only slightly decreased the number of SA-β-gal-positive cells, was sufficient to convert a miR146a-positive phenotype into a miR146a-negative one (Fig. 5C).Fig. 5. Characterization of the SASP profile and miR146a promoter activation in TIS A549 and MCF-7 cancer cells.A Qualitative and quantitative composition of the extracellular secretome in the cell-free supernatants of A549 and MCF-7 cells was characterized using multiplex bead-based immunoassays on day 7 across proliferative/untreated and TIS phenotypes. Each cell in the heatmaps represents the measured concentration of a single analyte (row) in the indicated proliferative/TIS condition (column). Color scale encodes the log₂ transformation of raw abundances. B Figure illustrates representative flow cytometry plots showing the gating of miR146a-eGFP+ cells of proliferative (untreated) and TIS phenotypes. The histogram shows the miR146a-eGFP MFI (fold-change) as a function of treatment condition (mean ± SD, n = 3; 0: proliferative; 1: palbociclib TIS; 2: doxorubicin TIS; 3: alisertib TIS; 4: bleomycin TIS). Data represent the mean ± SD of three independent experiments. Statistically significant differences between proliferative and TIS phenotypes are shown (* p < 0.05; **p < 0.005) using ANOVA analysis. C Representative images (n = 3) of SA-β-gal staining in A549 lung cancer cells were taken after treatment with 20 μmol/L bleomycin in the absence or presence of graded concentrations of etomoxir (ETO, 100 and 200 μmol/L) for 7 days (scale bar: 200 μm). Qualitative and quantitative composition of the extracellular secretome in cell-free supernatants of bleomycin with or without ETO and miR146a-eGFP expression was measured and represented as described in (A, B).
Together, these data decouple the SASP abundance from its inflammatory state. While the overall cytokine landscapes were similar and lineage-imprinted, the TIS stressors differentially engaged an NF-κB/miR146a-positive inflammatory module. Thus, miR146a sensing indicates a stressor-specific, mitochondria/FAO-dependent inflammatory SASP axis that correlates with senolytic responsiveness.
Generating senolysis-resistant senescent cancer cells requires preventing the generation of SASP downstream of mitochondria
Inflachromene (ICM) is a direct HMGB1/2 binder that blocks their cytoplasmic translocation and extracellular release. In primary fibroblasts, it induces a rapid, replicative-like senescence program while transcriptionally and proteomically constraining the pro-inflammatory SASP and its paracrine effects [60–63]. Mechanistically, ICM targets HMGB1/2-dependent chromatin/RNA hubs, which normally conduit inflammatory NF-κB signaling to activate the full SASP. Based on previous studies in fibroblasts and the potential mechanistic value of HMGB inhibition in uncoupling the SASP from senescence [60], we here used ICM for the first time in epithelial cancer cells to determine if mitochondrial metabolic reprogramming and senolytic responses are dispensable when an inflammatory SASP is directly prevented in the nucleus of senescent cancer cells.
First, we confirmed that ICM treatment induced a bona fide senescent phenotype in A549 and MCF-7 cells. These cells exhibited profound cytomorphological remodeling similar to that observed with the well-established TIS inducers palbociclib, doxorubicin, alisertib, and bleomycin (Fig. S6). This remodeling included an enlarged, flattened cell shape and a vacuole-rich cytoplasm, accompanied by SA-β-gal positivity (Fig. 6A). Markers of cell cycle arrest, such as loss of phospho-RB, downregulation of PCNA, and accumulation of p21^WAF1/Cip1^, occurred in ICM-induced senescent A549 and MCF-7 cells to a similar extent as with the classical TIS inducers palbociclib, doxorubicin, alisertib, and bleomycin (Fig. S1A, B). The canonical senescent phenotype induced by ICM was accompanied by a poor activation of the baseline SASP profile already present in A549 cells. While this was less evident in MCF-7 cells, the ICM-induced SASP was still not comparable to that observed with bleomycin (Fig. 6B). Accordingly, activation of the miR146a promoter was nearly absent in ICM-induced A549 senescent cells and slightly higher but still low in ICM-induced MCF-7 senescent cells (Fig. 6B).Fig. 6. Phenotyping of mitochondrial function, SASP profiling, miR146a promoter activation, and senolytic response in inflachromene (ICM)-induced senescent cancer cells.A Senescent phenotype of ICM TIS cancer cells. Top. Representative microphotographs (n = 3) showing the cytomorphological remodeling of ICM TIS cancer cells using the Incucyte S3 Adherent Cell-by-Cell Analysis System in phase contrast. Bottom. Representative images (n = 3) of SA-β-gal staining were taken after treatment with ICM for 7 days (scale bar: 200 μm) with drug refreshment on day 3 for A549 cells at 40 μmol/L and for MCF-7 cells at 20 μmol/L. B Top. Qualitative and quantitative composition of the extracellular secretome in the cell-free supernatants of proliferative/untreated and ICM TIS A549 and MCF-7 cells was characterized using multiplex bead-based immunoassays on day 7. Each cell in the heatmaps represents the measured concentration of a single analyte (row) in the indicated proliferative/ICM TIS condition (column). Color scale encodes the log₂ transformation of raw abundances. Bottom. Representative flow cytometry plots showing the gating of miR146a-eGFP+ cells of proliferative (untreated) and ICM TIS phenotypes. The histogram shows the miR146a-eGFP MFI (fold-change) as a function of treatment condition (mean ± SD, n = 3). C Left. Representative images (n = 3) of green mitochondrial staining with MitoView™ in live proliferative and ICM TIS cancer cells (scale bar: 500 μm). Right. Fold-change in mitochondrial content (FMC) was measured by flow cytometry as described in the “Materials and methods” section. Data represent the mean ± SD of three independent experiments performed in triplicate. Statistically significant differences between proliferative and TIS phenotypes are shown (* p < 0.05; **p < 0.005) using ANOVA analysis. D Heat maps of the AUC values for Biolog Mitoplate™ S-1-based measurements of the utilization of 31 metabolites by ICM TIS A549 and MCF-7 cancer cells over 1–6 h. Red indicates higher utilization. The mean AUCs are shown for n ≥ 3 independent experiments performed in triplicate. E Bar graphs show the ABT-263/navitoclax and A1331852 senolytic indexes obtained by dividing the IC_50_ values of ABT-263/navitoclax in proliferative A549 and MCF-7 cells by the values obtained in their ICM TIS derivatives. Data represent the mean ± SD of three independent experiments performed in triplicate.
After confirming that ICM induces a canonical senescent phenotype largely devoid of SASP in cancer cells, we investigated whether mitochondrial mass and/or the mitochondrial bioenergetic phenotype change in ICM-induced senescent cancer cells. We found that mitochondrial mass increased 2- to 3-fold in ICM-induced A549 and MCF-7 senescent cells (Fig. 6C). Furthermore, mitochondrial functional phenotyping once again revealed distinct mitochondrial substrate utilization profiles in ICM-induced senescent cells compared to their proliferative parental counterparts (Fig. 6D). In particular, ICM-induced A549 senescent cells exhibited a significant rewiring of mitochondrial bioenergetics. This rewiring involved not only an increased global bioenergetic output but also a substantial increase in the utilization of numerous metabolic sources for energy production. These sources include β-oxidation, a process that was also observed in the more poorly activated ICM-induced MCF-7 senescent cells (Fig. S7). Both ICM-induced A549 and MCF-7 senescent cancer cells exhibited a similar decrease in BCL2 expression of accompanied by a comparable BCL2L1 (BCL-xL/xS) upregulation and significantly increased CPT1B expression (Fig. S9). Interestingly, when tested for their response to ABT-263/navitoclax and A1331852, the ICM-induced A549 and MCF-7 senescent cancer cells behaved similarly to their proliferative parental counterparts in that they were fully resistant to senolysis (Figs. 6E and S9).
ICM uncouples senescence from the inflammatory SASP while preserving robust mitochondrial expansion and bioenergetic rewiring. Interestingly, even highly bioenergetic A549 cells are completely refractory to BCL-xL senolytics under these SASP-null conditions. These results imply that a high baseline bioenergetic (and primed) status and further mitochondrial bioenergetic rewiring alone are insufficient for senolysis and that a fully developed mitochondria-to-nucleus SASP axis is required.
Discussion
Mitochondria are increasingly recognized as integrators of senescence fate decisions, rather than merely as passive bioenergetic organelles [4–7]. In TIS, this role is clinically significant because senescent cancer cells can serve as a reservoir of residual disease or as an intentional byproduct that enables “one-two punch” senogenic-senolytic regimens [96]. This study addresses the central question of whether the functional bioenergetic state of TIS mitochondria—specifically, their flexible capacity to generate electron flow in the ETC from diverse NADH/FADH_2_-producing fuels—meaningfully governs vulnerability to BH3-mimetic senolytics. Additionally, the study explores whether mitochondrial control of the SASP is a required downstream node for senolysis.
TIS in cancer cells was accompanied by significant yet non-uniform changes in global mitochondrial activity and substrate utilization range across mechanistically distinct senogenic drugs that represent replicative, genotoxic, mitotic, and oxidative stress. Total bioenergetic capacity and substrate diversity exhibited graded patterns that varied by cell-of-origin. These results suggest that mitochondrial metabolic flexibility is not a stereotyped hallmark of TIS, but rather an emergent property of the specific senogenic insults depending on the cellular context. Indeed, pathway-resolved patterns support a model in which different senescence programs recruit different mitochondrial fuel circuits. In A549 cells, bleomycin TIS showed a disproportionate increase in malate-coupled substrates, carnitine-linked fatty acid β-oxidation (FAO) fuels, amino acid catabolism, and malate-aspartate shuttle intermediates. In contrast, alisertib TIS preferentially increased upper glycolytic/PPP-linked electron donation. The overutilization of pyruvate and citrate conserved across TIS phenotypes implies a shared elevation of glycolysis-to-TCA entry and early TCA flux. Meanwhile, phenotype-specific preferences (e.g., acetyl- and octanoyl-carnitine in bleomycin TIS) reveal stress-tuned rewiring of substrate entry points. Notably, the recurrent use of tryptamine across TIS phenotypes suggests the previously unappreciated involvement of MAO-linked redox fueling in senescent cancer cells [97–101]. Qualitatively, these “mitophenotypes” aligned with the magnitude of senolysis toward TIS cancer cells. In A549 cells, palbociclib TIS had the least flexible and least energetic mitochondria and was nearly refractory to ABT-263/navitoclax and A1331852. In contrast, bleomycin TIS had the most expanded substrate range and energetic output and was the most sensitive. Therefore, mitochondrial bioenergetic flexibility appears to reliably fine-tune the level of senolytic response in BH3 mimetic-sensitive TIS cancer cells. This is consistent with the notion that an energetically remodeled senescent mitochondriome is closer to an apoptotic “cliff edge” that BH3 mimetics can exploit.
One of the major conceptual advances of our findings is that the quantitative senolytic response, operationalized as senolytic indexes, does not increase proportionally with the extent of mitochondrial metabolic flexibility promoted by TIS stressors within each cell line. In fact, the bioenergetic configuration (as well as apoptotic priming) of the parental, pre-senescent mitochondria largely limits senolytic indexes. This constraint was evident when comparing lineages with different basal mitochondrial wiring. A549 cells, which are intrinsically highly primed and bioenergetically competent [48, 49, 51], converted stress-encoded flexibility into very large senolytic indexes (>100 for bleomycin TIS with A1331852). MCF-7 cells, which have limited BH3 dependence and lower basal bioenergetic competence, displayed a significant stress-dependent mitochondrial remodeling yet achieved only modest senolytic indexes (∼2–3 at best). At the other extreme, BAX-mutant LoVo cells, which are unprimed and with the poorest baseline bioenergetic capacity [41, 48, 49, 51], failed to become senolytic despite acquiring a senescent phenotype and showing some metabolic shifts in their TIS mitochondria. Succinate oxidation provides a particularly clean functional readout of this inherited ceiling. Since succinate donates electrons directly to Complex II and the CoQ pool, its oxidation bypasses the transport of upstream substrates and the variability of the TCA cycle, reporting the organization and stability of the ETC itself. In A549 and MCF-7 cells, which have a strong baseline of Complex II-CoQ flux, succinate utilization increased further upon senescence. In contrast, in LoVo cells, which have a weak baseline of Complex II capacity, the oxidation of succinate decreased significantly in the senescent state. Baseline succinate oxidation and the broader pre-TIS bioenergetic fingerprint (A549» > MCF-7>LoVo) correspond with the sensitivity of the derived TIS phenotypes to senolytics. Taken together, these data suggest that the concept of an inherent mitochondrial apoptotic priming extends to the bioenergetic domain [48–52]. TIS-associated bioenergetic remodeling can reposition mitochondria along an apoptotic landscape. However, the height of this landscape is inherited. In other words, the maximal amplitude of the senolytic response is determined by the mitochondrial “heritage” from the pre-senescent parental cells, while the TIS stress-specific mitophenotype shifts the amplitude of the senolytic response between minimum and maximum values. Future studies should evaluate the mechanistic link between the inherent capacity of the mitochondrial respiratory chain complex to adapt dynamically to the metabolic requirements of senescent cells and apoptotic priming [102].
SASP analyses revealed a mechanistic link between mitochondrial metabolic flexibility and senolytic sensitivity. Although TIS inducers with distinct mechanisms generated similar SASP profiles, which were largely predetermined by the baseline secretome of the cell-of-origin [41], activation of an NF-κB-responsive miR-146a promoter clearly distinguished senolytic responses from resistant phenotypes. This promoter was used as a functional SASP inflammation sensor. In both A549 and MCF-7 models, bleomycin, alisertib, and doxorubicin TIS were positive for miR-146a, whereas palbociclib TIS was negative. These results suggest that inflammatory SASP output, rather than SASP abundance per se, is the relevant downstream state for senolysis. The bioenergetic features of the miR-146a-positive mitophenotypes suggest a metabolic route to the establishment of inflammatory SASP. These mitophenotypes exhibited coordinated transcriptional upregulation of carnitine shuttle and FAO genes (e.g., CPT2 and SLC25A20 in A549 cells and CPT1B and SLC25A1 in MCF-7 cells), corresponding with the observed increase in FAO-linked substrate oxidation. Since FAO generates high levels of NADH/FADH_2_ and produces mitochondrial acetyl-CoA, this shift could support ETC hyperactivity and the production of signaling metabolites, such as citrate and acetyl-carnitine. These metabolites can then feed nuclear acetyl-CoA pools and promote histone acetylation and SASP gene transcription [81–83]. Consistent with this coupling, the pharmacological inhibition of FAO with etomoxir converted bleomycin TIS from miR-146a-positive to miR-146a-negative, despite only modestly affecting the fraction of SA-β-gal-positive cells. Therefore, mitochondrial substrate routing in TIS cancer cells is not merely an energetic adaptation, but rather, it is a lever that controls inflammatory SASP licensing.
The second key finding emerged from the ICM experiments. The ICM is a dual HMGB1/2 chromatin-hub inhibitor [60] that induces a bona fide senescent phenotype in A549 and MCF-7 cancer cells while producing a largely SASP-null, miR146a-negative state. Notably, ICM-induced senescent cancer cells still increased mitochondrial mass and exhibited a clear and, in the case of A549 cells, strong bioenergetic reprogramming. However, despite their inherited and further potentiated bioenergetically active and highly flexible mitochondrial state, ICM-induced senescent cells were fully resistant to ABT-263 and A1331852. They behaved similarly to their proliferative counterparts in response to BH3 pro-apoptotic stimuli. These results suggest that the establishment of an inflammatory SASP is not merely an incidental consequence of mitochondrial remodeling, but rather a necessary prerequisite for senolytic killing. This finding is mechanistically consistent with HMGB biology: HMGB2 maintains SASP loci in an open chromatin configuration and prevents SAHF-mediated silencing, while HMGB1 organizes 3D chromatin and RNA availability for SASP transcripts [61–63]. By blocking HMGB1/2, ICM interrupts the mitochondrial retrograde signaling axis at the level of nuclear SASP competence. This produces a senescent yet senolysis-refractory state.
Functional mapping of mitochondrial bioenergetic flexibility provides a predictive approach to one-two punch senogenic-senolytic strategies. The MitoPlate-derived mitophenotype, specifically the metrics of succinate oxidation and substrate diversity, can categorize cancer cells according to their responsiveness to senolytics before senescence is induced. ICM results suggest that senolytic failure may be caused not only by inadequate priming and/or bioenergetic remodeling, but also by suppressed inflammatory SASP programs, an important consideration as anti-inflammatory senomorphics enter oncology [103–105]. However, several limitations temper the scope of the conclusions. First, the cell line panel is narrow (including A549, MCF-7, and a BAX-mutant LoVo extreme), necessitating broader sampling across tissue origins and genetic backgrounds to generalize the “mitochondrial heritage” ceiling and refine baseline fingerprints that predict senolytic magnitude. Second, the MitoPlate platform provides high-content functional phenotyping of mitochondrial electron flow, yet it does not directly resolve causality for specific pathways. More targeted substrate tracing and pathway perturbations are required to substantiate the proposed mechanistic link between flexibility, acetyl-CoA flux, and inflammatory SASP licensing. Examples of these perturbations include the controlled provision of FAO versus glycolytic substrates and the genetic manipulation of CPT1/2, MPC, or CoQ redox regulators. Third, although miR-146a activation is a powerful integrative reporter of NF-kB-driven inflammatory SASP, it does not identify which individual SASP factors are necessary or sufficient for senolysis. Dissecting these components, including potential non-canonical mitochondrial DAMP outputs [3, 106], remains an open task. Finally, these findings are derived from in vitro TIS models. In vivo validation is essential to determine how stromal interactions, immune surveillance, and therapeutic pharmacokinetics shape the mitochondria-SASP-senolysis circuit in “one-two punch” regimens and other physiological and pathological scenarios, including normal tissues [107].
Conclusions
This work establishes mitochondrial bioenergetic flexibility as a stress-encoded factor that fine-tunes the degree of senolytic intensity. However, it reveals that the extent of the senolytic scale largely depends on the mitochondrial bioenergetic configuration of the parental cell (Fig. 7). First-in-class ICM experiments in cancer cells demonstrate most notably that inflammatory SASP competence downstream of mitochondrial signaling is required for BH3-mediated senolysis, even in highly primed, bioenergetically active senescent cancer cells. This dual-layer model—mitochondrial bioenergetic heritage setting the ceiling and SASP licensing enabling execution—offers a more predictive and mechanistically grounded framework for deploying senolytics in cancer therapy.Fig. 7. Mitochondrial bioenergetics-SASP cross-talk: A framework to understand senolytic sensitivity of TIS cancer cells.Our data support a layered control architecture for senolysis in TIS cancer cells. 1. Response ceiling: The maximum achievable senolytic index is limited by the bioenergetic (and priming) traits of the parental mitochondrial lineage. These traits can be captured via baseline succinate/CII oxidation and the global bioenergetic output of pre-senescent mitochondria. 2. Gradient switching: Within each response ceiling, the degree of mitochondrial flexibility encoded by the TIS stressor fine-tunes the ultimate magnitude of the senolytic response. The greater the global bioenergetic generation from a higher number of usable mitochondrial metabolites, the greater the sensitivity to senolysis by BCL-xL-targeting BH3 mimetics. Low-flexibility mitochondrial states (e.g., palbociclib TIS) are too far from apoptotic commitment, while high-flexibility states are closer to the cliff edge. 3. Licensing gate: Even when mitochondrial flexibility (and priming) is permissive, a mitochondria-driven inflammatory SASP (miR-146a-positive and competent for NF-κB) is required to translate bioenergetic and apoptotic stress into BH3-sensitized death. In its absence, senescence becomes a senolysis-resistant terminal state. Mitochondrial bioenergetic remodeling can generate substrate-driven inflammatory signaling and anti-apoptotic dependencies. BH3 mimetics can only exploit these dependencies if the inflammatory SASP axis is engaged. (FAO: fatty acid oxidation).
Materials and methods
Cell lines and culture
A549 (ATCC CCL-185), MCF-7 (ATCC HTB-22), LoVo (ATCC CCL-229), and HEK293T (ATCC CRL-3216) cell lines were obtained from the ATCC (Manassas, VA, USA). Cells were routinely expanded in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Linus), 1% L-glutamine, 1% sodium pyruvate, 50 IU/mL penicillin, and 50 μg/mL streptomycin. Cells were grown at 37 °C in a humidified atmosphere containing 5% CO_2_ and were in the logarithmic growth phase at the beginning of the experiments. Cell lines were authenticated by STR profiling, both performed by the manufacturer and confirmed in-house at the time of purchase according to ATCC guidelines. Cells were passaged by starting a low-passage cell stock every month until to 2–3 months after resuscitation. Cell lines were screened for mycoplasma contamination using a PCR-based method for Mycoplasma detection prior to experimentation and were intermittently tested thereafter.
Drugs and reagents
Doxorubicin (#S1208), PD-0332991/palbociclib (#S1116), ABT-263/navitoclax (#S1001), A1331852 (Cat. #S7801), etomoxir (Cat. #S8244), and an antibody against PCNA (Cat. #F0018) were purchased from Selleckchem (Houston, TX, USA). MLN8237/alisertib (#331-10890-1) and bleomycin sulfate (#331-11727-3) were purchased from RayBiotech Inc. (Norcross, GA, USA). Inflachromene (#CAY-17006) was purchased from Cayman Chemical (Ann Arbor, MI, USA). AlamarBlue™ Cell Viability Reagent (Cat. #DAL1100), PageRuler™ Plus Prestained Protein Ladder (Cat. #26619), and 7-AAD (7-aminoactinomicin D; Cat. #A1310) were purchased from Thermo Fisher Scientific Inc. (Waltham, MA, USA). The senescence β-galactosidase staining kit (Cat. #9860) and antibody against p21^WAF1/CIP1^ (12D1; Cat. #2947) were purchased from Cell Signaling Technology, Inc. (Danvers, MA, USA). Antibodies against phospho-RB1 (Ser807/811; Cat. #30376-1-AP), β-actin (#66009-1-Ig) and GAPDH (#60004-1-Ig) were purchased from Proteintech Group, Inc. (Rosemont, IL, USA).
MitoView™ Green was purchased from Biotium, Inc. (Fremont, CA, USA). The PHAGE-PmiR-146a-GFP-PGK-puro plasmid [58] was kindly provided by Stephen Elledge (Department of Genetics, Harvard Medical School, Division of Genetics, Brigham and Women’s Hospital, Howard Hughes Medical Institute, Boston, MA, USA). The lentivirus packaging system consisting of pCMV-VSV-G (#8454) and pCMV-dR9.2 dvpr (#8455) was purchased from Addgene (Cambridge, MA, USA).
Therapy-induced senescence
We identified the optimal initial cell densities and senescence-inducing concentrations that allowed A549 and MCF-7 cell cultures to reach approximately 60–80% confluence after 7 days of treatment without drug refreshment. These concentrations were 5 μmol/L palbociclib, 50 nmol/L doxorubicin, 500 nmol/L alisertib, and 20 μmol/L bleomycin [51]. Inflachromene treatment lasted also for 7 days with drug refreshment on day 3 for A549 cells at 40 μmol/L and for MCF-7 cells at 20 μmol/L. These concentrations and regimens promoted the appearance of the major classical markers of senescence, namely enlarged and flattened cell shape and increased senescence-associated β-galactosidase (SA-β-gal) activity in the highest number of cells. Cellular β-galactosidase activity was detected using the Senescence β-Galactosidase Staining Kit according to the manufacturer’s instructions.
Phase contrast imaging
Cells were imaged using a 10× objective on the Incucyte S3 Adherent Cell-by-Cell Analysis System (Sartorius) in phase contrast. The images were processed using IncuCyte software, version 2022B.
Cell cycle distribution
Cell cycle distribution was analyzed using 7-AAD on a CytoFLEX SRT Cell Sorter (Beckman Coulter), and the data were analyzed using Kaluza Analysis software.
Immunoblotting
Immunoblotting, including cell lysis, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), transfer, and incubation with primary and secondary antibodies, was performed as previously described [51].
Mitochondrial function phenotyping
Mitochondrial activity in proliferative and TIS cancer cells was measured in triplicate using 96-well MitoPlate S-1 plates (Biolog). Wells containing the various cytoplasmic and mitochondrial metabolic substrates (n = 31) were rehydrated with a solution containing mitochondrial assay solution (MAS), redox dye mix, and 30 μg/mL saponin in sterile water. Cells were washed with PBS, resuspended in 1× MAS, and added to each well at a final cell density of 30,000 cells/well. Mitochondrial substrate metabolism was assessed by monitoring the colorimetric change of the terminal electron acceptor tetrazolium redox dye at a wavelength of 590 nm on a kinetic microplate reader (0 to 6 h).
To quantitatively compare the mitochondrial metabolism of proliferative cells and TIS phenotypes, we developed a scoring system based on the slope ratios (area under the curve, AUC) of the absorbance values of the mitochondrial substrates normalized to the negative control wells. Each substrate was scored separately, and mitochondrial metabolism scores were generated based on either the normalized AUCs of all substrates or functionally related metabolites (i.e., glycolysis/PPP, TCA, ETC, G3P, MAS, β-oxidation, amino acids) combined.
Senolytic index
Cell viability was measured using alamarBlue™ assays. The non-toxic, cell-permeable alamarBlue™ (resazurin) reagent is an oxidized form of a redox indicator that is blue in color and non-fluorescent. Upon entering living cells, the reagent is reduced to resorufin, which changes color from blue to red and becomes fluorescent. The oxidized environment upon loss of cell viability maintains the non-fluorescent, blue color of resazurin.
To calculate senolytic indexes, TIS cancer cells and their proliferative counterparts were harvested after 7 days of treatment with TIS inducers, reseeded in 96-well plates (3000 cells/100 μL/well proliferative A549 cells vs 6000 cells/100 μL/well TIS A549 cells; 4000 cells/100 μL/well proliferative MCF-7 cells vs 8000 cells/100 μL/well TIS MCF-7 cells) and exposed to graded concentrations of ABT-263/navitoclax and A1331852 for an additional 5 days. Cells were incubated with the alamarBlue solution (10 μL/well) at 37 °C for 4 h, and the increase in fluorescence signal was measured using a fluorescence detector. The senolytic index was defined as the ratio between the inhibitory concentration 50 (IC_50_) values of proliferative cells and the IC_50_ values of bleomycin-, alisertib-, doxorubicin-, palbociclib-, and inflachromene-induced TIS cancer cells. The IC_50_ values were defined as the concentration of drug that produced a 50% reduction in control fluorescence (by interpolation).
Mitochondria quantification
Cells were seeded in 100-mm cell culture dishes at a density of 4 × 105 cells/plate. After 24 h, the cells were treated for 7 days as described above. After treatment, the culture medium was replaced with prewarmed complete medium containing 100 nmol/L MitoView™ Green and cells were incubated for 30 min under normal culture conditions. The medium was then changed to live cell imaging solution, and cells were imaged with 20×/40× objective on a live cell microscope (Nikon Eclipse Ts2) equipped with a MicroScopia Digital XM Full HD camera using a 470 LED filter. Finally, cell samples were collected by trypsinization, resuspended in PBS-5% FBS and analyzed using a BD Accuri C6 flow cytometer (BD Biosciences, San Jose, CA, USA). The FITC mean fluorescence intensity (MFI) of each condition was measured using an identical gating strategy for all experiments, and collecting at least 30,000 live, single, and non-dead cell events for each condition. Data from two independent experiments were presented by plotting the FITC MFI of each treatment, and the fold mitochondrial content (FMC) was calculated by dividing the MFI of each TIS condition by the MFI of the proliferative control cells.
RNA isolation and reverse transcription
Total RNA was extracted from cells using RNA Plus Kit (Macherey-Nagel, Germany) following the manufacturer’s instructions. Two micrograms of the total RNA were then reverse-transcribed into cDNA using a High-Capacity cDNA Reverse Transcription Kit (Life Technologies, Carlsbad, CA, USA). The concentration and quality of the RNA were determined using a NanoDrop™ ND-1000 spectrophotometer (NanoDrop Technologies, USA).
Gene expression assays (Custom Array Plates)
Gene expression was assessed using quantitative Real-Time PCR with TaqMan Custom Array Plates (Life Technologies) containing primers and fluorescent probes for the following genes: ACLY, CPT1B, CPT2, CS, BCL2, BCL2L1, ETFDH, MPC1, NDUFAB1, SDHA, SHMT2, SLC25A1, and SLC25A20 (TaqMan Gene Expression Assay IDs: Hs00982738_m1, Hs03046298_s1, Hs00988962_m1, Hs02574374_s1, Hs04986394_s1, Hs00236329_m1, Hs01031780_m1, Hs00211484_m1, Hs00900741_g1, Hs07291714_mH, Hs01059263_g1, Hs01105608_g1, and Hs00386383_m1, respectively). The 18 s rRNA (Hs99999901_s1), PPIA (Hs99999904_m1), and RPLP0 (Hs99999902_m1) genes were used as reference genes.
Secretome
The qualitative and quantitative composition of TIS-associated secretomes in terms of cytokines, chemokines, and growth factors was characterized using multiplex bead-based immunoassays, Luminex®. Briefly, cell culture supernatants from non-senescent (proliferative controls) and cancer cells rendered senescent with mechanistically different TIS inducers as described above were collected in strict parallel. Collected extracellular milieus were sent to a third-party commercial laboratory (Eve Technologies, Canada) to perform the Panel A Cytokine/Chemokine 48-Plex Discovery Assay® Array (HD48A) in a Luminex® platform.
Lentiviral transduction and miR146a-EGFP flow cytometry
Viral particles were generated in 293T cells by co-transfecting the PHAGE-PmiR-146a-GFP-PGK-puro plasmid with the pCMV-VSV-G/ pCMV-dR8.2 dvpr 3rd generation lentivirus packaging system. 293T cells were transiently transfected to generate lentiviral supernatants, and A549 and MCF-7 cells were infected with lentiviral supernatants at 8 μg/mL Polybrene. After 48 h, medium containing 10 μg/mL puromycin was added for another 48 h to generate miR146a-EGFP reporter-containing A549 cells.
To measure miR146a induction as a surrogate marker of response to SASP, miR146A-eGFP reporter-containing A549/MCF-7 cells were rendered senescent as described above, harvested, and resuspended in 300 μL of medium. eGFP positivity was determined using a laser excitation wavelength of 488 nm (FITC channel) on a BD Accuri C6 flow cytometer (BD Biosciences, San Jose, CA, USA). Data were analyzed using FCS Express 7 software (De Novo™ Software, Pasadena, CA, USA) and are representative mean fluorescence intensities (arbitrary units) from three independent experiments.
Statistical analysis
All cell-based observations were confirmed by at least three independent experiments performed in triplicate for each cell line and for each condition. Data are expressed as mean ± SEM or ± SD. Bar graphs, curves, and statistical analyses were generated using GraphPad Prism 10 (GraphPad Software, San Diego, CA). Two-group comparisons were performed using Student’s t test for paired and unpaired values. Comparisons of means of ≥3 groups were performed by ANOVA, and the existence of individual differences, in the case of significant F values in ANOVA, was tested by Dunnett’s multiple contrasts. p < 0.05 was considered to be statistically significant. All statistical tests were two-tailed.
Supplementary information
Supplementary Material Uncropped Western Blots
The reference list from the paper itself. Each links out to its DOI / PubMed record.