Hypothesis

The aging brain actively evicts metabolically inefficient neurons via microglial phagoptosis, redirecting scarce energy resources to high‑yield circuits. This process is not random decay but a calibrated response to declining ATP supply, mediated by CD36‑dependent recognition of neuronal “inefficiency flags” such as low firing rates, mitochondrial ROS, and reduced BDNF/TrkB signaling. When neuronal energy demand falls below a tissue‑specific threshold, microglia upregulate CD36, bind exposed phosphatidylserine or oxidized lipids, and engulf the neuron before it dies, thereby preserving network efficiency per watt.

Mechanistic Extension

We propose that the inefficiency signal is amplified by a neuronal‑microglial metabolic checkpoint: neuronal SIRT1 activity declines with age, leading to hyperacetylation of PGC‑1α and reduced mitochondrial biogenesis. Low SIRT1 also diminishes deacetylation of the neuronal surface protein CX3CR1, increasing its affinity for microglial CD36. Simultaneously, age‑related rise in extracellular ADP activates microglial P2Y12 receptors, priming them for phagocytic uptake. Together, these changes create a feed‑forward loop where falling SIRT1 → higher CX3CR1‑CD36 binding → phagoptosis → further SIRT1 loss in neighboring neurons due to disrupted neurotrophic support.

Testable Predictions

1. Neuronal SIRT1 levels predict phagoptosis susceptibility. In aged mice, neurons with SIRT1 expression below the 20th percentile will show increased colocalization with microglial CD36 and phagocytic cups (measured by imaging flow cytometry).
2. Blocking CD36 rescues inefficient neurons without restoring global metabolism. Antibody‑mediated CD36 inhibition in middle‑aged mice will increase the number of low‑firing, low‑BDNF neurons (identified by c‑Fos and BDNF immunoreactivity) while ATP levels remain unchanged.
3. Artificial elevation of neuronal SIRT1 attenuates phagoptosis. Viral overexpression of SIRT1 in hippocampal excitatory neurons will reduce CD36‑positive microglial contacts and improve memory performance despite unchanged global cerebral glucose metabolism (FDG‑PET).
4. Microglial P2Y12 antagonism disrupts the efficiency checkpoint. Treatment with a selective P2Y12 antagonist will decrease phagoptosis markers (Iba1+CD68+ phagocytic cups) and increase the variance of neuronal firing rates, indicating loss of selective pruning.

Experimental Approach

• Use Cre‑dependent SIRT1‑overexpressing AAV in Camk2a‑Cre mice; assess neuronal survival with stereological NeuN counts and phagoptosis via electron microscopy of microglial‑neuronal contacts.
• Apply in vivo two‑photon imaging of CX3CR1‑GFP microglia and tdTomato neurons to track real‑time engulfment events after CD36 blockade.
• Measure extracellular ADP with biosensors; correlate with microglial activation state (phospho‑ERK) and phagocytic index.
• Perform seahorse flux analysis on isolated neurons sorted by SIRT1 level to confirm metabolic inefficiency correlates with phagoptosis risk.

Falsifiability

If CD36 blockade fails to increase the survival of low‑SIRT1 neurons, or if SIRT1 overexpression does not reduce phagoptosis despite restoring mitochondrial function, the hypothesis that a neuronal‑microglial metabolic checkpoint drives selective eviction would be refuted. Likewise, demonstrating that phagoptosis occurs independently of SIRT1/CX3CR1‑CD36 interactions would falsify the proposed mechanistic link.

Implications

Reframing age‑related neuronal loss as an adaptive, albeit maladaptive when excessive, response shifts therapeutic goals from preventing cell death to modulating the efficiency checkpoint—e.g., fine‑tuning microglial CD36 activity or bolstering neuronal SIRT1 to preserve useful networks without compromising the brain’s energy economy.