Engineered Flavonoid Cousins of Fisetin Clear Senescent Cells and Boost Healthspan in Mice

Senescent cells—the dormant, damaged cells that pile up in our tissues as we age—have long been considered one of the fundamental drivers of aging itself. Now, a team of researchers has engineered two new flavonoid-based drugs that clear these cells far more effectively than fisetin, the plant-derived compound that has become one of the most closely watched senolytic candidates in the field. In a study published in Aging Cell, the scientists describe how a phenotypic drug discovery strategy, guided by structure–activity relationship analysis, produced two optimized analogs, SR29384 and SR31133, that reduced senescence markers across multiple tissues in aged mice and improved selected measures of healthspan in a mouse model of accelerated aging.

The rationale for the work begins with a paradox. Fisetin, a naturally occurring flavonoid found in strawberries and other plants, has shown promise in preclinical studies, reducing senescence markers, alleviating inflammation, and extending healthspan in aged mice. Those findings have fueled multiple clinical trials testing fisetin in chronic kidney disease, osteoarthritis, skeletal health, COVID-19, and frailty. Yet fisetin is, by the authors’ own measurements, not a strong senolytic. It requires high doses in cell culture to kill senescent cells, and it suffers from poor solubility and absorption, properties that could limit its clinical effectiveness. The team set out to fix these weaknesses without abandoning the flavonoid scaffold entirely.

Because fisetin is promiscuous—activating or suppressing a wide range of pathways involved in antioxidant defense, apoptosis, proliferation, inflammation, and angiogenesis—traditional structure-based drug design, which depends on a predefined protein target, was poorly suited to the problem. Instead, the researchers turned to phenotypic drug discovery, screening compounds for their ability to change a visible cellular phenotype: the activity of senescence-associated beta-galactosidase, a classic marker of senescent cells. Using a high-content fluorescent imaging platform, they screened a focused library of natural flavonoids spanning flavones, flavonols, flavanols, anthocyanidins, and other subclasses in senescent mouse embryonic fibroblasts lacking the DNA repair protein ERCC1, alongside matched non-senescent controls.

The screening results were revealing. Fisetin showed only moderate senolytic activity, with an EC50 of roughly 22.74 micromolar and a low selectivity index of 1.53, meaning it was barely more toxic to senescent cells than to healthy proliferating ones. Luteolin, a flavone found in many vegetables, emerged as the most potent and selective natural flavonoid in the screen, with an EC50 of 14.21 micromolar. From these data, the team extracted preliminary structure–activity relationship rules: planar scaffolds retaining the C2–C3 double bond and C4-oxo group of flavones and flavonols were consistently active, while isoflavones, flavanones, and anthocyanidins were largely inert. A 3′,4′-dihydroxyl pattern on the B ring proved essential, a 5-hydroxyl on the A ring hurt activity, electron-donating methoxy groups helped, and any glycosylation was detrimental.

Armed with these guidelines, the researchers launched an iterative cycle of design, synthesis, and biological testing, ultimately creating 35 novel flavonoid analogs. Most outperformed both fisetin and luteolin. Substituents at the 6-position of the A ring proved more favorable than those at the 7-position, and bulky groups such as tert-butyl and isopropyl markedly enhanced activity. Two leads emerged. SR29384, bearing an isopropyl substituent, was the most selective compound in the entire series. SR31133, carrying a five-membered alicyclic ring, was the most potent, with an EC50 of 0.8 micromolar—a more than 27-fold improvement in senolytic activity over fisetin. Both analogs also showed improved computed drug-like properties, including greater lipophilicity, lower topological polar surface area consistent with better blood–brain barrier permeability, and reduced predicted toxicity and carcinogenicity, though the authors caution that such in silico predictions require empirical longitudinal validation.

The optimized analogs proved to be broad-spectrum senolytics. They reduced the number of senescent cells across a diverse panel of models, including Ercc1-deficient fibroblasts, wild-type fibroblasts stressed by oxidative or genotoxic damage, senescent human IMR90 lung fibroblasts, and replicatively senescent human umbilical vein endothelial cells. Treatment also significantly downregulated key senescence and inflammatory markers, including p16Ink4a, p21Cip1, and interleukin-6, outperforming fisetin. Notably, a DPPH radical scavenging assay showed that all three compounds had comparable antioxidant activity, indicating that the enhanced senolytic power of the new analogs was not simply a matter of better antioxidant chemistry but reflected genuinely altered interactions with cellular targets.

The in vivo results were the most striking. In 28-month-old wild-type C57BL/6J mice—an age at which fisetin would not be expected to be significantly effective—the team administered a deliberately low dose of 20 milligrams per kilogram per day for five consecutive days, a stringent comparison against the 100 milligrams per kilogram used in earlier fisetin studies. Fisetin had only marginal effects on senescence and SASP markers. The two analogs, particularly SR29384, significantly decreased expression of p16Ink4a and p21Cip1 as well as multiple SASP factors, including Il6, Il1β, Mcp1, and Pai1, in tissues including kidney, brain, lung, spleen, fat, and muscle. The systemic reduction of senescence markers in diverse organs, including the brain, served as functional evidence that the compounds possess sufficient oral bioavailability to engage their targets, even though direct pharmacokinetic measurements were not performed.

SR29384 was then tested in Ercc1-deficient progeroid mice, a model of accelerated aging. Treated three times per week for five weeks at 20 milligrams per kilogram by oral gavage, the compound suppressed the composite score of aging symptoms, with particularly notable improvements in dystonia and ataxia, two indicators of aging-related neurodegeneration. Gene expression analysis showed significant reductions or trends toward reduction in senescence and SASP factors across the kidney, liver, lung, and brain. The authors note that the effects were tissue- and marker-dependent, and that SR29384 did not show uniform superiority across all tissues, an honest caveat that underscores the complexity of translating senolytic activity from cell culture to living organisms.

Perhaps the most scientifically valuable contribution of the study is its mechanistic dissection. By integrating bulk RNA sequencing, machine-learning-based pathway prediction, structure-based target prediction, protein–protein interaction network analysis, and molecular docking, the team converged on PARP1 and CDK2 as the key shared targets of both analogs, with BCL-xL implicated as a possible third. Docking simulations suggested that both analogs bind the PARP1 catalytic cavity more strongly than fisetin, which suffered unfavorable donor–donor clashes with key residues, and that they compete with ATP in the CDK2 activation domain. Enzyme assays confirmed these predictions: SR31133 inhibited PARP1 with an IC50 of 2.51 micromolar, outcompeting the standard inhibitor 3-AB, and inhibited CDK2 with an IC50 of 12.73 micromolar. Neither analog directly inhibited BCL-xL at the concentrations tested, suggesting that the observed perturbation of BCL-xL signaling networks occurs indirectly. Supporting the functional relevance of these targets, the PARP1 inhibitor olaparib reduced SASP gene expression and modestly reduced SA-β-gal positivity in senescent IMR90 cells, while the CDK2 inhibitor staurosporine reduced SASP genes and p16Ink4a expression.

The proposed mechanism is a multi-hit assault on the pro-survival machinery of senescent cells. Senescent cells resist apoptosis through upregulated anti-death pathways, and the new analogs appear to exploit this vulnerability by disrupting DNA damage detection through PARP1 inhibition, interfering with cell cycle and stress signaling through CDK2, and potentially undermining BCL-xL-mediated survival through crosstalk among these proteins. The two compounds also showed distinct secondary profiles: SR29384’s exclusive targets were enriched for inflammatory signaling and extracellular matrix remodeling, which may explain its remarkable selectivity, while SR31133’s exclusive targets pointed toward histone deacetylation processes. The authors emphasize that formal pharmacokinetic, tissue-distribution, and target-engagement studies are needed next, and that the compounds remain preclinical. Still, by pairing a phenotypic screening platform with modern computational biology, the study demonstrates how an aging-related natural product with murky mechanisms can be transformed into rationally optimized drug candidates—and brings the field a step closer to senotherapeutics that could one day treat multiple age-related diseases at once.

Subject of Research: Development of novel flavonoid senolytic drugs that clear senescent cells and improve healthspan markers in mouse models of aging

Article Title: Novel Flavonoid Senotherapeutics Identified by Phenotypic Drug Discovery Reduce Senescence and Improve Multiple Markers of Healthspan

Article References: Zhang, L. J., Salekeen, R., Soto‐Palma, C., Wu, Q., He, Y., Elsallabi, O., Bdiri, B., Hughes, B., Nunes, A., Xu, W., Zhang, B., Mohamed, A., McGowan, S. J., Angelini, L., O'Kelly, R., Kamenecka, T. M., Niedernhofer, L. J., & Robbins, P. D. (2026). Novel Flavonoid Senotherapeutics Identified by Phenotypic Drug Discovery Reduce Senescence and Improve Multiple Markers of Healthspan. Aging Cell, 25(10), Article e70736. https://doi.org/10.1111/acel.70736

Image Credits: AI Generated

DOI: 10.1111/acel.70736

Keywords: senolytics, cellular senescence, fisetin, flavonoids, SASP, PARP1, CDK2, phenotypic drug discovery, healthspan, aging, Aging Cell, preclinical