Cellular senescence has moved from the periphery of aging biology to one of its most intensively studied mechanisms in the last decade. The basic observation is now well-established: cells that have undergone replicative exhaustion, oncogenic stress, or DNA damage can enter a stable non-proliferative state characterized by altered morphology, resistance to apoptosis, and a complex secretory profile that affects surrounding tissue. As the organism ages, the fraction of senescent cells increases across tissues, and the cumulative effect of their secretome appears to drive chronic low-grade inflammation, tissue dysfunction, and impaired regeneration.
The therapeutic question is whether anything can be done about this accumulation. Two strategies have been explored — clearing senescent cells (senolytics) or suppressing their secretory phenotype (senomorphics) — and both have generated enough preclinical data to attract serious pharmaceutical interest. The less-explored question we are thinking about at caVos is whether mRNA-based approaches could offer a third option: reprogramming the senescent state itself.
This article is about hypotheses and open questions. We do not have validated mRNA approaches to cellular senescence. We have a framework for thinking about which targets might be designable as mRNA therapeutics, and a set of observations about where the evidence supports or fails to support the intervention logic.
What Senescent Cells Actually Do: SASP in Detail
The senescence-associated secretory phenotype (SASP) is not a single entity. It is a complex, dynamic secretory profile that varies by cell type, senescence-inducing stimulus, and time since senescence onset. The core components include pro-inflammatory cytokines (IL-6, IL-8, TNF-alpha), matrix metalloproteinases (MMP-1, MMP-3, MMP-10), growth factors, and in some contexts, extracellular vesicles carrying damage signals.
The SASP is regulated by multiple transcription factors. NF-kappaB is probably the dominant driver, but C/EBP-beta, AP-1, and mTORC1 signaling all contribute to its activation and maintenance. The SASP is also not uniform over time: early-phase SASP is more immunostimulatory (potentially beneficial for oncogene-induced senescence surveillance), while chronic SASP in aging tissues has a more damaging paracrine character.
p16/INK4a and p21/CIP1 are the canonical markers used to identify senescent cells. p16 (encoded by CDKN2A) suppresses CDK4/6 and maintains Rb in its growth-suppressive hypophosphorylated state, locking the cell in G1. p21 (encoded by CDKN1A) is also a CDK inhibitor but is induced by a broader range of stresses, including DNA damage, and is less specific to the stable senescent state. Cells expressing high p16 with irreversible cell cycle arrest and SASP activation are the most canonically senescent; p21-positive cells may be transiently arrested and capable of re-entry into the cell cycle.
This heterogeneity matters for any therapeutic intervention: "senescent cells" is not a uniform population. A treatment that effectively targets p16-high senescent fibroblasts may have entirely different effects on p21-high progenitor cells under stress.
The Senolytic Landscape: Where Things Stand
Senolytics — compounds that selectively kill senescent cells — have advanced considerably in the past decade. The most studied combination is dasatinib (a BCR-ABL/Src kinase inhibitor) plus quercetin (a flavonoid), which was identified through transcriptomic analysis of senescent cell survival pathways. The logic is that senescent cells upregulate specific anti-apoptotic programs (Bcl-2 family, HSP90, PI3K/AKT pathway components) to resist their own apoptosis signaling. Drugs that suppress these survival programs preferentially kill senescent cells relative to proliferating cells.
Navitoclax, a Bcl-2/Bcl-xL inhibitor, is another senolytic with a clearer mechanism but significant hematological toxicity (thrombocytopenia) that has limited its clinical development outside oncology. Unity Biotechnology has explored intraarticular injection for knee osteoarthritis to limit systemic exposure. ABT-737 and related compounds are active in preclinical models.
The preclinical data in rodent models for senolytics is genuinely compelling — multiple independent groups have shown lifespan extension and healthspan improvement in mouse models using intermittent senolytic dosing. The critical open question is human translation. Rodent senescent cell accumulation kinetics, SASP composition, and tissue context differ from humans. The clinical trials in older adults that have been reported are early and small, with endpoints focused on safety and senescent cell biomarkers rather than disease outcomes.
Senomorphics — agents that suppress SASP without killing senescent cells — include rapamycin (mTORC1 inhibition, which reduces SASP), metformin, and various JAK inhibitors. These have different risk profiles from senolytics and may be more appropriate in contexts where senescent cell clearance could remove beneficial populations (for example, senescent cells that support wound healing through SASP-mediated fibroblast recruitment).
Can mRNA Approaches Target Senescence?
This is where we move from established biology into hypothesis space, and we want to be clear about that transition. The following represents our current thinking about what might be designable, not what has been demonstrated.
Hypothesis 1: mRNA Encoding Apoptosis-Triggering Proteins Selectively in Senescent Cells
One approach would be to deliver an mRNA encoding a pro-apoptotic protein whose expression is controlled by a senescence-responsive regulatory element. The idea is that the mRNA would be translated preferentially in senescent cells due to transcriptional or post-transcriptional elements that are active in those cells but not in healthy neighbors.
The challenge is specificity. There is no single transcription factor or RNA-binding protein that is uniquely active in senescent cells across all cell types. p16 promoter elements have been used to drive senescent-specific transgene expression in lentiviral constructs in preclinical models (the CAR-T senolytic approach published by the Bhanu Bhanu Bhanu Bhanu — we are referring here to work from the Bhanu Bhanu approach — note: we are not citing a specific paper, we are describing a conceptual approach that has been reported in the literature using p16-promoter-driven constructs). But promoter-driven selectivity requires stable transgene integration, which is not the mRNA modality.
For mRNA specifically, post-transcriptional control via miRNA-responsive elements in the 3' UTR is a more tractable route. Senescent cells have an altered miRNA expression profile. If miRNAs that are low or absent in senescent cells (but present in healthy cells) can be used as inverse de-repression switches, you could design an mRNA that is translated only when those miRNAs are absent — i.e., in senescent cells. This is a real engineering principle used in cell-type-specific mRNA delivery, and the question is whether the senescence-specific miRNA signature is consistent enough to serve as a reliable switch. We do not know the answer to that question yet, and it would require a significant amount of single-cell miRNA profiling in aged human tissue to even formulate the design properly.
Hypothesis 2: mRNA Encoding SASP-Suppressing Regulators
An alternative to killing senescent cells is suppressing their secretome. mRNA encoding dominant-negative NF-kappaB components, overexpression of IkB (the endogenous NF-kappaB inhibitor), or delivery of regulators that compete with SASP-activating transcription factors could reduce the inflammatory impact without requiring cell death.
This approach has a cleaner safety profile conceptually — you are not engineering cell death, you are modulating a signaling pathway. The limitation is that it is not genuinely senolytic; you are managing the consequence of senescence rather than addressing the cell population. Whether repeated mRNA dosing to maintain NF-kappaB suppression in senescent cells would be practical is a significant pharmacological question.
Hypothesis 3: mRNA-Mediated Partial Reprogramming
The most ambitious hypothesis in the senescence field is whether it is possible to reverse the senescent state using reprogramming factors. The Yamanaka factors (Oct4, Sox2, Klf4, c-Myc — OSKM) can reset epigenetic age marks when expressed transiently in differentiated cells without full dedifferentiation. Several groups have shown partial reprogramming in mouse models that reduced markers of epigenetic aging. The question is whether senescent cells are amenable to this kind of reset, and whether mRNA delivery of reprogramming factors is a viable route.
mRNA has an inherent advantage here: transient expression profile, no genomic integration, controlled dosing. Several companies are exploring mRNA-delivered partial reprogramming, including work on Yamanaka factor mRNA constructs. The evidence that this approach is effective at the organismal level in mammals beyond rodents is not established. There are significant concerns about incomplete reprogramming producing partially dedifferentiated cells with genomic instability — the safety profile of partial reprogramming in aged tissue is genuinely unknown.
We are not pursuing partial reprogramming as a primary approach at caVos. Our focus is on upregulating longevity-associated proteins whose conservation across long-lived species suggests a favorable benefit-risk profile. But the senescence reprogramming hypothesis is worth tracking because it represents the most mechanistically direct approach to the epigenetic component of cellular aging.
Evidence Gaps That Matter
Several things would need to be true for mRNA-based senescence intervention to be practical, and the evidence for them ranges from weak to absent:
- Consistent senescence signature across tissues: The SASP composition and marker expression in liver senescent cells, lung senescent cells, and neuronal senescent cells are not identical. A single-target mRNA approach may work in one tissue context and fail in another.
- Delivery to senescent cell populations: LNP tropism is currently strong for liver after systemic administration and depends on LNP formulation for CNS. Reaching senescent fibroblasts in skin, cartilage, or other peripheral tissues at therapeutic concentrations without off-target delivery is a formulation problem that is not solved.
- Effect size relative to background: Even if an mRNA construct successfully kills or quiets senescent cells in a target tissue, the fraction of total cells that are senescent in any given tissue is typically less than 10% even in aged organisms. Whether reducing that fraction by, say, 50% produces a measurable functional improvement in humans is an empirical question that has not been answered.
Why We Are Watching This Space Without Committing to It Yet
caVos is focused on longevity-associated protein targets identified through our comparative genomics pipeline. FOXO3, Klotho, and related proteins appear in our prioritization because they score well on cross-species conservation, positive selection analysis, and pathway connectivity. Several of these targets have indirect connections to senescence biology — FOXO3 suppresses NF-kappaB and has pro-apoptotic activity that could influence senescent cell survival; Klotho affects IGF-1/PI3K signaling which has been implicated in SASP regulation.
So the relationship between our primary approach and senescence intervention is not accidental. But we are not claiming that our current pipeline addresses cellular senescence directly. The evidence for senescence as a druggable target in humans is more advanced than for many longevity biology hypotheses, which is why it deserves serious attention. Whether mRNA is the right modality for that intervention, or whether small-molecule senolytics will get there first, is not clear. The field will benefit from parallel development of multiple modalities rather than premature convergence on one approach.
The open questions in this space are genuinely open: how to achieve selective delivery to senescent populations, what the appropriate markers of target engagement are for a senolytic mRNA in vivo, and whether the benefit seen in aged rodent models will translate to the more complex senescent landscape in aged human tissue. These are hard problems. The mRNA modality has specific properties — transient expression, tuneable dose, no genomic integration — that could be advantages for senescence intervention, but those advantages only matter if the fundamental selectivity problem is solved first.
