Metabolic Checkpoint Hypothesis: Aging Neurons Are Tagged for Microglial Phagocytosis by ATP‑Dependent 'Eat Me' Signals, Not Random Damage

Hypothesis

Aging neurons are selectively removed not by indiscriminate damage but through a regulated metabolic checkpoint that flags low‑energy cells for microglial phagocytosis. This checkpoint couples intracellular ATP/AMP ratio and mitochondrial membrane potential to surface exposure of phosphatidylserine (PS) and downregulation of CD47, thereby converting a metabolic stress signal into an 'eat me' cue. When systemic inflammation chronically elevates complement C1q, the phagocytic signal is amplified, leading to excessive neuronal loss that mirrors synaptic pruning pathology but operates at the level of whole cells.

Mechanistic Basis

1. Energy Sensing – Neurons with sustained AMPK activation and low NAD+ levels exhibit reduced mitochondrial membrane potential (Δψm). Low Δψm promotes PS externalization via scramblase activation, a well‑characterized apoptosis‑independent 'eat me' signal【https://pmc.ncbi.nlm.nih.gov/articles/PMC3710114/】.
2. Immune Modulation – PS‑exposed neurons bind microglial TIM‑4 and BAI1, triggering phagocytosis. Concurrently, age‑related decline in neuronal CD47 expression diminishes the 'don't eat me' signal, shifting the balance toward engulfment【https://www.bu.edu/kilachandcenter/cognitive-decline-in-old-age-may-be-linked-to-increased-pruning-of-brain-cell-connections/】.
3. Complement Amplification – Elevated C1q in the aged brain opsonizes PS‑rich neurons, enhancing microglial uptake via CR3 receptors. Blocking C1q reduces phagocytosis without affecting baseline synaptic pruning, indicating a separable pathway【https://www.pnas.org/doi/10.1073/pnas.2010281117】.
4. Regional Vulnerability – High‑metabolism, insulin‑sensitive regions (prefrontal cortex, hippocampus) experience earlier NAD+ decline and insulin resistance, predisposing them to the metabolic checkpoint【https://www.pnas.org/doi/10.1073/pnas.2416433122】【https://www.jci.org/articles/view/158453】.

Testable Predictions

• Neurons with experimentally reduced ATP (e.g., via rotenone) will show increased surface PS and decreased CD47, correlating with higher microglial engulfment in vitro.
• Genetic overexpression of neuronal CD47 or supplementation with NAD+ precursors will resist age‑related neuronal loss in mouse models, even when C1q is elevated.
• Pharmacological inhibition of AMPK will prevent PS exposure and reduce neuronal eviction, without altering synaptic density.
• In aged C1q‑deficient mice, neuronal loss will be attenuated despite persistent metabolic stress, whereas synaptic pruning remains unchanged.

Experimental Approach

1. In vitro – Primary cortical neurons treated with metabolic stressors (oligomycin, 2‑DG) will be stained for Annexin V (PS) and CD47, then co‑cultured with microglial BV2 lines; phagocytosis quantified via flow cytometry.
2. In vivo – Aged WT and C1q‑KO mice receive either NAD+ booster (NR) or AAV‑mediated neuronal CD47 overexpression; stereological neuron counts in hippocampus and prefrontal cortex assessed after 3 months.
3. Intervention – AMPK inhibitor (Compound C) delivered via intracerebroventricular pump; outcomes measured for neuronal survival, synaptic markers (Synapsin‑1), and behavioral performance (Morris water maze).
4. Readouts – Mitochondrial membrane potential (TMRM), NAD+ levels (enzymatic assay), cytokine profiling (ELISA for C1q, IL‑1β), and microglial activation state (Iba1, CD68).

If these experiments confirm that boosting neuronal energy status or blocking the 'eat me' signal rescues neurons without affecting synapse numbers, the hypothesis that aging brain employs a metabolic quality‑control mechanism—rather than passive damage—will be supported. Conversely, failure to rescue neurons under these conditions would falsify the model, reinforcing the view that age‑related neuronal loss is primarily a consequence of irreversible cellular damage.