Hypothesis

Aging neurons are actively removed when their mitochondrial NAD+‑dependent oxidative capacity falls below a threshold, triggering increased surface C1q exposure and decreased CD47 signaling, which together direct microglia to phagocytose the inefficient cells.

Mechanistic Rationale

• Synaptic pruning in adulthood relies on activity‑dependent complement tagging (C1q) and protective CD47 signals [3].
• Neuronal survival correlates with firing‑driven metabolic demand; low‑activity synapses are preferentially eliminated [4].
• Mitochondrial NAD+ levels decline with age, reducing ATP production and altering the redox state of neurons [5].
• Low NAD+ activates SIRT1‑dependent deacetylation of NF‑κB, increasing C1q transcription and decreasing CD47 expression on the neuronal surface [6].
• Microglia recognize the altered C1q/CD47 ratio via CR3 and SIRPα receptors, leading to phagocytic uptake [2].

Thus, neuronal eviction is not random damage but a metabolic checkpoint that couples energetic inefficiency to immune‑mediated clearance.

Testable Predictions

1. Elevating neuronal NAD+ in aged brains (via NMN supplementation or NAMPT overexpression) will reduce C1q upregulation and increase CD47 preservation, decreasing microglial phagocytosis of neurons.
2. Neuron‑specific knockdown of NAMPT will accelerate C1q rise, CD47 loss, and synaptic/neuronal loss, even in young adult mice.
3. Blocking neuronal C1q release (using antisense oligonucleotides) will protect NAD‑deficient neurons from microglial engulfment without altering their metabolic state.
4. In vivo two‑photon imaging will show a temporal sequence: NAD+ decline → increased C1q surface labeling → decreased CD47 → microglial process extension and neuronal uptake.

Experimental Approach

• Use Cre‑dependent NAMPT floxed mice crossed with CamKII‑Cre to delete NAMPT in forebrain excitatory neurons; assess NAD+ levels, C1q/CD47 surface staining (flow cytometry), microglial phagocytic markers (Iba1+/CD68+), neuron numbers (NeuN+), and cognitive performance (Morris water maze) at 3, 9, and 18 months.
• Parallel cohort receives chronic NMN (400 mg/kg/day) in drinking water; compare to vehicle controls.
• Apply anti‑C1q antibodies or neuron‑targeted siRNA to test necessity of the complement signal.
• Employ intravital imaging of thy1‑YFP mice with NAD+ biosensor (SoNar) and fluorescently labeled microglia to visualize the proposed sequence.

Falsifiability

If NAD+ restoration fails to modify C1q/CD47 balance or neuronal survival, or if manipulating C1q/CD47 does not affect phagocytosis despite NAD+ changes, the hypothesis would be refuted, indicating that neuronal eviction in aging operates independently of this metabolic‑immune checkpoint.