Hypothesis

The aging brain does not merely prune underperforming synapses; it actively eliminates whole neurons that fall below a critical ATP‑dependent viability threshold. This somatic eviction mirrors synaptic pruning logic—microglia recognize and phagocytose neurons exposing “eat‑me” signals when mitochondrial output cannot sustain basal ionic homeostasis.

Mechanistic Basis

• Metabolic checkpoint: Neurons continuously gauge their ATP/ADP ratio via AMPK. When ATP drops persistently, AMPK phosphorylates the lipid scramblase XKR8, driving phosphatidylserine (PS) exposure on the somatic surface.
• Microglial recognition: PS, together with upregulated complement C1q binding to stressed neurites, serves as a dual‑signal that tags the neuron for microglial engulfment, akin to the C1q‑mediated synapse elimination pathway described in 1 and 2.
• Energy budget coupling: In aging, cerebral glucose metabolism declines and mitochondrial efficiency wanes, lowering the average ATP yield per neuron. Neurons with high firing rates or poor axonal transport become the first to breach the threshold, aligning with the observation that metabolically expensive, weakly connected cells are preferentially lost.
• Homeostatic vs. pathological regimes: Under normal aging, the checkpoint triggers a low‑grade, compensatory eviction that preserves network efficiency by removing the most energetically costly nodes. Chronic inflammation or amyloid‑β toxicity can hyperactivate the pathway, causing excessive neuronal loss as seen in neurodegeneration 3.

Testable Predictions

1. PS exposure correlates with neuronal loss – In aged mouse hippocampus, neurons exhibiting somatic PS staining (detected by annexin‑V‑based live imaging) will be significantly more likely to be absent after a 4‑week chase, whereas PS‑negative neighbors survive.
2. AMPK deficiency blunts eviction – Neuron‑specific AMPK knockout mice will retain higher neuronal counts in cortical layers II/III despite showing comparable synaptic pruning (measured by C1q‑coated synapse density). Behavioral assays will reveal preserved baseline cognition but impaired ability to suppress hyperexcitability during metabolic stress.
3. Boosting neuronal ATP rescues soma without affecting synapses – Chronic nicotinamide riboside supplementation will raise neuronal NAD⁺/ATP levels, reducing somatic PS exposure and microglial neuron phagocytosis by ~40% while leaving synaptic pruning rates unchanged, as quantified by synaptic C1q puncta.
4. Complement dependence – Administration of a C1q‑blocking antibody will decrease microglial engulfment of somal targets in aged mice, confirming that the same complement tagging that marks weak synapses also flags energy‑deficient neurons.

Potential Implications

If validated, this hypothesis reframes age‑related neuronal loss as a programmed quality‑control mechanism rather than indiscriminate decay. Therapeutic strategies could then aim to reset the metabolic checkpoint (e.g., via NAD⁺ boosters or AMPK modulators) to preserve neuronal numbers without disrupting beneficial synaptic refinement. Conversely, chronic checkpoint activation may underlie cognitive decline in neurodegenerative diseases, suggesting that complement‑or PS‑targeted interventions could mitigate pathogenic neuronal eviction while sparing adaptive synaptic pruning.