A First Somatic Cell Aging Model for Systems Beyond

A Boolean network for proliferation, apoptosis, and senescence

1. Why we built this model

Aging doesn’t happen in one organelle or one pathway. It emerges from how nutrient sensing, stress responses, DNA damage, and inflammatory signals talk to each other over time. To even start asking serious “what if?” questions (CR, rapamycin, NAD⁺, anti-inflammatories, etc.), we need a generalizable somatic cell model, not tied to a specific tissue or disease.

The model presented here is our first computational prototype of such a cell within Systems Beyond:

• It’s mechanistic enough to include key players (mTOR, AMPK, ROS, DDR, p53, NFκB, etc.).

• It’s simple enough (Boolean logic) to explore whole-network dynamics, attractors, and basins of attraction.

• It can already express three essential fate outcomes: proliferation, apoptosis, and senescence + SASP.

This is Model v1.0: a starting point for systematic refinement, not a final “truth”.

2. Model structure in a nutshell

The model is a Boolean network with:

• 8 external inputs (environment / context): NUTRIENTS, GF, DNA_DAMAGE, ER_STRESS, INFLAM_ENV, HYPOXIA, MATRIX, OX_STRESS

• 42 internal nodes, grouped into:

  • Receptors & signaling: RTK, INTEGRIN, SRC, SYK, LAT, PLC, PI3K, PDK1, AKT, PKC, RAS, AP1, NFAT, NFKB, …

  • Metabolic core: LKB1, AMPK, SIRTUIN, MTOR, MTORC1, MTORC2, HIF1A, GLYCOLYSIS, MITO_FUNC, OXPHOS

  • Stress / damage / checkpoints: ROS, NRF2, FOXO, AUTOPHAGY, DDR, P53, P21, P16_RB, BCL2

  • Cell fates: APOPTOSIS, SENESCENCE, SASP, PROLIFERATION

Each node is updated synchronously using logic rules like:

AMPK	= LKB1 and ((not NUTRIENTS) or ER_STRESS or ROS or HYPOXIA)
SIRTUIN	= AMPK and not NUTRIENTS
MTORC1	= MTOR and not AMPK
ROS	= OX_STRESS or (not MITO_FUNC) or (INFLAM_ENV and NFKB)
DDR	= DNA_DAMAGE or (ROS and not AUTOPHAGY)
P53	= DDR
SENESCENCE	= (P53 and (DDR or ROS) and MTORC1) or (SENESCENCE and (DDR or ROS or INFLAM_ENV))
APOPTOSIS	= (P53 and (DDR or ROS) and not BCL2 and not MTORC2) or (APOPTOSIS and P53)
PROLIFERATION	= RTK and INTEGRIN and MTORC1 and GLYCOLYSIS and not P21 and not P16_RB and not APOPTOSIS and not SENESCENCE

So:

• Nutrient & growth factor signals drive AKT → MTORC1 → GLYCOLYSIS → PROLIFERATION.

• Energy & stress sensors (LKB1/AMPK/SIRTUIN) oppose MTORC1 and promote FOXO and AUTOPHAGY.

• Damage accumulation (DNA_DAMAGE, ROS) feeds into DDR → P53 → P21/P16 and then either:

  • Apoptosis (if survival wiring via BCL2 is weak and MTORC2 is low), or

  • Senescence (if MTORC1 is high and damage/inflammation persist).

NFκB and SASP sit at the inflammaging interface: senescent cells plus NFκB activation drive an inflammatory secretory phenotype that can, in turn, increase local ROS and damage.

3. How the model behaves: three example scenarios

We tested the model in three simple environments, using the Jupyter code you now have.

Scenario 1 – Proliferative, low-stress environment

Inputs:

• Nutrient rich, growth factor present: NUTRIENTS = 1, GF = 1

• No external damage or inflammation: DNA_DAMAGE = ER_STRESS = INFLAM_ENV = OX_STRESS = 0

• Well attached, normoxic: MATRIX = 1, HYPOXIA = 0

What happens:

• RTK and INTEGRIN turn on → AKT activates.

• AMPK stays off (no energy stress), so MTORC1 is active.

• HIF1A (via MTORC1) + nutrients → GLYCOLYSIS = 1.

• No DDR → P53 = P21 = P16_RB = 0.

Attractor:

• PROLIFERATION = 1

• APOPTOSIS = 0, SENESCENCE = 0, SASP = 0

• ROS low, AUTOPHAGY off or low, SIRTUIN mostly off.

Interpretation: this is a young, healthy, proliferative cell in a favorable environment. The attractor is stable and easy to reach from many random initial internal states.

Scenario 2 – Chronic damage + inflammatory environment (“aging-like”)

Inputs:

• Nutrients and growth factor still present: NUTRIENTS = GF = 1

• Persistent damage & stress: DNA_DAMAGE = ER_STRESS = INFLAM_ENV = OX_STRESS = 1

• Matrix and oxygen OK: MATRIX = 1, HYPOXIA = 0

What happens:

• ROS and DDR remain chronically high.

• P53 activates (driven by DDR).

• P21 and P16_RB turn on and stay on.

• AMPK and SIRTUIN see stress but are constantly opposed by nutrient / GF signaling and active AKT/MTOR.

Depending on the precise internal starting state, the system tends to move away from clean proliferation and into cell-cycle arrest with elevated stress, often favoring APOPTOSIS as the eventual attractor:

• PROLIFERATION usually collapses (blocked by P21/P16_RB and damage).

• APOPTOSIS turns on when P53 is high, DDR/ROS are present, and BCL2 is weak.

• SENESCENCE can appear if MTORC1 remains high at the moment of strong P53 + damage, but with these harsh conditions apoptosis often dominates.

Interpretation: this is closer to “damage overload” than clean senescence. The network tends to choose “die” rather than “stay alive but arrested,” which is biologically plausible at high damage.

Scenario 3 – Entry into senescence

To explicitly demonstrate senescence as a distinct attractor, we used:

Inputs:

• Nutrients available, but no growth factor:NUTRIENTS = 1, GF = 0

• Attached, some oxidative stress:MATRIX = 1, OX_STRESS = 1

• No strong inflammation or ER stress initially:INFLAM_ENV = ER_STRESS = 0, HYPOXIA = 0

Internal “pre-damaged” state:

We initialize the internal nodes in a configuration representing a proliferating, already damaged but not yet senescent cell:

• DDR = 1, P21 = 1, P16_RB = 1, but SENESCENCE = 0

• Survival and mitochondrial function are still present (MITO_FUNC = 1, OXPHOS = 1)

• Mild signaling and NFκB activity via PKC and ROS

• PROLIFERATION = 1 initially (the cell is still trying to cycle despite checkpoints)

When we iterate the Boolean dynamics from this state:

• DDR and ROS feed into P53.

• SENESCENCE rule sees P53 + damage + some MTORC1 and flips to 1.

• Once SENESCENCE = 1, the feedback part of the rule (SENESCENCE and (DDR or ROS or INFLAM_ENV)) keeps it on.

• SASP turns on because SENESCENCE = 1 and NFKB is active.

The system settles into an attractor with:

• SENESCENCE = 1

• SASP = 1

• PROLIFERATION = 0

• APOPTOSIS = 0 (this is a live, non-dividing cell)

• Sustained stress markers (DDR, ROS, P53) plus inflammatory output.

Interpretation: this is a senescent, SASP-positive cell — a good first approximation of an aging cell that is alive but permanently arrested and broadcasting inflammatory signals to its environment.

4. What this model already tells us

Even at this Boolean resolution, a few patterns emerge:

1. Proliferation requires a tight window. You need:

   • Nutrients + growth factor (RTK, INTEGRIN, MTORC1, GLYCOLYSIS),

   • Low checkpoints (P21 = P16_RB = 0),

   • And absence of strong fate commitments (APOPTOSIS = SENESCENCE = 0).Once damage pushes P53 and P21/P16_RB up, the cell easily falls out of this window.

2. Senescence is a “compromise attractor.” It appears when:

   • Damage (DDR, ROS) and P53 are high,

   • mTORC1 remains active (nutrient/growth context),

   • Survival signaling (BCL2, mitochondrial function) is not fully collapsed.This fits the idea that senescence is more likely in nutrient-replete, growth-signaling environments where the cell is too damaged to divide but too supported to die.

3. Apoptosis vs. senescence is tunable. In code form, the balance between:

   • MTORC1 (growth push) and

   • BCL2 / AKT vs P53 / ROS / DDRdecides whether cells die or become senescent. This is a natural axis for in silico experiments on therapies that:

   • Push damaged cells toward apoptosis, or

   • Suppress SASP without removing senescent cells.

4. NFκB is central for inflammaging.

   • NFκB is activated by PKC, ROS, and INFLAM_ENV.

   • It feeds into ROS and is required for SASP.Blocking NFκB in this model can give you senescent but SASP-low cells — which mimics the idea of “non-inflammatory senescence” as a therapeutic goal.

5. SIRTUIN/AMPK axis shapes resilience but doesn’t magically “rejuvenate”.

   • Turning on AMPK and SIRTUIN pushes FOXO and AUTOPHAGY, and opposes MTORC1.

   • That reduces stress (ROS, DDR) and can make the apoptotic/senescent boundaries shift.But in a purely Boolean setting, once damage is fully “1” and senescence is stabilized, flipping AMPK/SIRTUIN on is usually not enough to reverse senescence. That’s a realistic constraint, not a bug: in real biology, full reversal is rare and context-dependent.

5. Limitations and next steps

This is intentionally a minimal, discrete model. Some limitations we already know:

• Binary states can’t capture graded responses (e.g. “low vs high” ROS, partial senescence, sublethal damage).

• Time scales are homogeneous — everything updates at the same step, while in reality transcriptional programs, protein activation, and damage accumulation occur on very different time horizons.

• Parameters are conceptual, not calibrated to specific cell types or data.

Planned next steps:

1. Systematic basin mapping.For each environment (e.g. high nutrients + mild EL damage vs severe), map which fraction of random initial conditions leads to proliferation, apoptosis, or senescence.

2. Multi-valued / fuzzy upgrade.Replace strict Booleans with 3–5 discrete levels (or sigmoidal ODEs) for key variables like ROS, DDR, MTORC1, AMPK, and SIRTUIN to better capture dose–response and partial rescue.

3. Data anchoring.Gradually align logic and states with experimental observations from:

   • Replicative senescence,

   • Oncogene-induced senescence,

   • CR / rapamycin / NAD⁺ interventions,

   • Chronic inflammatory exposure.

4. Interface to tissue-level models.Use SASP + NFκB + INFLAM_ENV as links into higher-level tissue models where senescent cells affect neighbors and immune cells.

6. How this fits into Systems Beyond

This somatic cell aging network is meant to be a shared reference model for the Beyond Aging project:

• A sandbox where hypotheses about mechanisms and interventions can be formalized and tested.

• A didactic object you can show in articles: network diagram + Boolean heatmaps + attractor statistics.

• A foundation for later continuous and multi-scale models (cells in tissues, organ-level dynamics, whole-organism trajectories).

From here you can:

• Publish this as a methods/first-results article (e.g. “A Boolean Somatic Cell Aging Model for Systems Beyond”).

• Invite collaborators to propose rule changes, new nodes, or data constraints.

• Start building a series of “virtual experiments” (CR, rapamycin, SIRT activators, NFκB inhibitors, senolytics) that you can run and compare systematically.

If you’d like, I can help you turn this into a more formal manuscript-style structure (abstract, methods, results, figures list, and a short discussion) ready to paste into your Systems Beyond site or a preprint.