Metformin Clears Aging Cartilage Cells in Lab Study by Unlocking Bak

Metformin, the widely used first-line drug for type 2 diabetes, can drive senescent human cartilage cells into apoptosis in the laboratory by physically rearranging the protein interactions that normally keep these aged cells alive, according to a new study published in Frontiers in Pharmacology. Using live-cell fluorescence resonance energy transfer (FRET) imaging in immortalized human chondrocytes, researchers at Shantou University Medical College found that metformin weakens the grip of the anti-apoptotic proteins Mcl-1 and Bcl-xL on the pro-apoptotic effector Bak, while simultaneously strengthening the association between Bak and the BH3-only activator Bim. The net result, they report, is Bak dimerization, loss of mitochondrial membrane potential, and caspase- and reactive oxygen species (ROS)-dependent cell death specifically measured in senescent chondrocytes.

The work addresses a long-standing question in osteoarthritis (OA) research. OA is a chronic, irreversible musculoskeletal disorder closely linked to aging, most commonly affecting the hip and knee joints, and there are currently no disease-modifying drugs for the condition. A key pathogenic factor is the accumulation of senescent chondrocytes beyond the body’s clearance capacity. These aged cells secrete a cocktail of inflammatory substances known as the senescence-associated secretory phenotype (SASP), which exacerbates pain and functional impairment in patients. Eliminating excessive senescent cells has therefore emerged as a potential therapeutic strategy, but the molecular switches that decide whether an aged cartilage cell lives or dies have remained incompletely mapped.

To build their experimental system, the team first induced senescence in immortalized human chondrocyte lines using the chemotherapy drug doxorubicin. After screening a low 50 nM dose, they observed that cell proliferation stopped within two to three days and the fraction of senescence-associated β-galactosidase positive cells surged from 2.5 percent to 79.8 percent. Western blotting confirmed the senescent state, with the cell-cycle arrest markers P16, P21, and P53 all elevated. This chemically induced aging model, the authors note, is a limitation, since it may not fully reproduce the replicative or stress-induced senescence seen in aging joints.

With the senescent model in hand, the researchers tested metformin’s effects on cell viability using CCK-8 assays. The drug inhibited senescent chondrocyte activity in a time- and concentration-dependent manner, with half-maximal inhibitory concentrations of 86.85 mM at 12 hours, 40.52 mM at 24 hours, and 17.36 mM at 48 hours. Normal chondrocytes were far more resistant at early time points, showing little toxicity at 100 mM within 12 hours, although their IC50 values dropped to 70.54 mM and 50.22 mM at 24 and 48 hours. Colony formation assays reinforced the selectivity signal: 40 mM metformin significantly inhibited senescent cell activity while leaving normal chondrocyte colony formation unaffected.

The death mechanism bore the biochemical hallmarks of classical apoptosis. JC-1 probe imaging showed that 40 mM metformin caused a significant loss of mitochondrial membrane potential in senescent cells, with no comparable effect on normal cells. Annexin-V/PI flow cytometry quantified a marked increase in apoptosis after 24 hours of treatment, and ROS measurements revealed a significant rise in intracellular reactive oxygen species that the antioxidant NAC could suppress. Western blotting detected activation of caspase-3, caspase-8, and caspase-9, along with cleavage of PARP, a canonical readout of caspase-3 activity. Critically, the pan-caspase inhibitor Z-VAD and the ROS scavenger NAC each significantly blocked metformin-induced senescent cell death, indicating that both caspase activity and oxidative stress are required for the killing process. The activation of caspase-8, the authors point out, hints at possible overlap between the extrinsic and intrinsic apoptotic pathways.

The study’s most technically distinctive contribution lies in its use of quantitative live-cell FRET to measure protein-protein interactions directly inside senescent cells. The Bcl-2 protein family governs mitochondrial apoptosis through a balance of anti-apoptotic sentinels and pro-apoptotic effectors: proteins such as Bcl-xL, Mcl-1, and Bcl-2 suppress cell death by binding and sequestering effectors like Bak and Bax, while BH3-only proteins such as Bim directly activate Bak to trigger mitochondrial outer membrane permeabilization. In the doxorubicin-induced senescent cells, FRET saturation assays across 89 cells co-expressing tagged Bcl-xL and Bak revealed a maximum FRET efficiency of 0.391 ± 0.021, substantially higher than the 0.203 ± 0.016 measured in control cells, indicating that aging tightens the Bcl-xL-Bak complex. A parallel analysis of 103 cells expressing Mcl-1 and Bak showed the same pattern, with senescent cells reaching an EDmax of 0.295 ± 0.027 versus 0.222 ± 0.020 in controls.

Metformin treatment reversed this sequestration. After drug exposure, the FRET efficiency for Bcl-xL/Bak fell to 0.142 ± 0.012 and for Mcl-1/Bak to 0.171 ± 0.015, both significantly below the senescent baseline, and co-immunoprecipitation experiments confirmed diminished Bak co-precipitation with both anti-apoptotic partners. At the same time, the drug reshaped Bcl-2 family expression: metformin upregulated Bak, Bax, and the BH3-only protein Bim while downregulating Mcl-1, Bcl-xL, and Bcl-2, and it reduced Puma expression. FRET measurements between Bim and Bak in 93 cells showed that metformin pushed their interaction to an EDmax of 0.410 ± 0.034, above both senescent (0.365 ± 0.0301) and control (0.276 ± 0.025) values. Protein cross-linking experiments then demonstrated that Bak dimers increased in senescent cells and increased further with metformin, consistent with a Bim-driven Bak activation model in which dimerization of Bak on the mitochondrial membrane opens the door to apoptotic factor release.

The authors are careful to flag what their data do and do not establish. Because metformin concurrently lowers Mcl-1 and Bcl-xL abundance while raising Bak levels, the reduced FRET signals may partly reflect altered protein expression rather than a genuine dissociation of pre-existing complexes, and quantitative normalized co-immunoprecipitation with functional rescue experiments would be needed to settle that question. Definitive proof that Bim is required for metformin-induced Bak oligomerization and apoptosis awaits genetic intervention studies such as siRNA knockdown or CRISPR knockout of Bim or Bak, and the possible contribution of Bax cannot be excluded without Bax-specific silencing. Moreover, while the CCK-8 assays showed differential toxicity between senescent and normal cells, all of the mechanistic apoptosis assays were performed exclusively in senescent cells, so the team cannot yet conclude that the identified molecular pathway is unique to aged chondrocytes.

Dose is another major caveat. The effective in vitro concentrations, roughly 30 to 40 mM, far exceed typical therapeutic plasma levels of metformin in patients, and the authors suggest that data at 20 mM and above may reflect non-specific stress responses rather than a clean senolytic mechanism. In a broader concentration sweep from 0.1 to 100 mM, senescent cells showed selective cytotoxicity within 24 hours at 20 mM, with apoptosis reaching 72.84 percent while normal chondrocyte apoptosis stayed at 3.11 percent, but whether such concentrations could be achieved safely in a joint through local or systemic delivery remains unknown and would require animal OA models such as DMM or ACLT for evaluation. The canonical AMPK pathway, through which metformin is widely believed to act, was not examined in this study, and future work must determine whether AMPK inhibition blunts or worsens the drug’s pro-apoptotic effect in senescent cells and whether the two pathways converge.

Within those boundaries, the study offers a proof-of-concept mechanistic account rather than a clinical result. The authors position metformin alongside other senolytic candidates under development, including dasatinib, quercetin, and Bcl-2 family inhibitors such as ABT-263 and A1331852, noting that ABT-263’s clinical use is limited by thrombocytopenia and neutropenia, whereas metformin carries decades of clinical experience and a well-characterized safety record at antidiabetic doses. By showing that metformin can, in cultured senescent human chondrocytes, break the inhibitory binding of Bcl-xL and Mcl-1 to Bak while promoting Bim-Bak activation and Bak dimerization, the work expands the known pharmacological repertoire of the drug and nominates the mitochondrial Bak axis as a potential intervention target for clearing aged cartilage cells. Translating that mechanism into a therapy for osteoarthritis, however, will depend on validation in primary chondrocytes, alternative senescence models, and in vivo joint delivery studies that have yet to be performed.

Subject of Research: Metformin-induced apoptosis of senescent human chondrocytes through Bcl-2 family protein interactions

Article Title: Metformin promotes apoptosis of senescent chondrocytes by disrupting the Bcl-xL/Mcl-1–Bak interaction

Article References: Feng, B., Liu, L., Lin, Y., & Wu, G. (2026). Metformin promotes apoptosis of senescent chondrocytes by disrupting the Bcl-xL/Mcl-1–Bak interaction. Frontiers in Pharmacology, 17, 1944776. https://doi.org/10.3389/fphar.2026.1944776

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1944776

Keywords: metformin, osteoarthritis, senescent chondrocytes, apoptosis, Bak, Bcl-2 family, Bim, Mcl-1, FRET, mitochondrial membrane potential, senolytics, cartilage