Core Idea

Autophagy acts as a cellular rationing system, but the choice of which components to consume is governed by a metabolic switch: the hepatic NAD+/SIRT3 axis. When fasting drives NAD+ above a critical level, SIRT3 deacetylates key autophagy adaptors, biasing the machinery toward lipid droplet lipophagy and sparing mitochondria. As NAD+ declines later in the fast, the same pathway shifts to favor mitophagy. This creates a bistable selector that explains the observed temporal hierarchy of selective autophagy.

Mechanistic Basis

• NAD+ rises steadily during fasting due to increased NAMPT activity and reduced consumption by PARPs and CD38.
• SIRT3, a mitochondrial sirtuin, is activated by NAD+‑dependent deacetylation. Its substrates include LC3‑interacting proteins (e.g., p62/SQSTM1) and phosphatidic acid phosphatase Lipin‑1.
• Acetylated p62 has lower affinity for ubiquitinated lipid droplets; deacetylated p62 preferentially binds PLIN2‑coated droplets, initiating lipophagy.
• Conversely, when NAD+ falls, SIRT3 activity drops, p62 remains acetylated, and its affinity for ubiquitinated mitochondrial proteins (e.g., MIRO1, TOM20) increases, steering autophagosomes toward mitophagy.
• This switch is reinforced by feedback: lipophagy‑derived acetyl‑CoA fuels cytosolic acetyl‑CoA synthetase 2 (ACSS2), which temporarily sustains nuclear acetylation and delays mitophagy until lipid stores are depleted.

Testable Predictions

1. Threshold NAD+ level: In mouse liver, NAD+ concentrations above ~500 nM correlate with a lipophagy/mitophagy ratio >2; below this threshold the ratio reverses.
2. SIRT3 dependence: SIRT3‑KO mice will show blunted lipophagy early in fasting (12‑20 h) and premature mitophagy, even when NAD+ is high.
3. Pharmacological shift: Acute NAD+ boosters (NR, NMN) administered at hour 16 will extend the lipophagy‑dominant window; SIRT3 inhibitors (3‑TYP) will compress it.
4. Readouts: Use mKeima‑LC3 constructs targeted to lipid droplets (PLIN2‑mKeima) and mitochondria (TOM20‑mKeima) to quantify organelle‑specific flux alongside NAD+ biosensors (SoNar) and SIRT3 activity assays.

Implications

If validated, this model reframes intermittent fasting benefits as a tunable rheostat: nutritional or pharmacological manipulation of NAD+/SIRT3 can bias autophagy toward lipid clearance without triggering excessive organelle loss. It also suggests that diseases with chronic NAD+ depletion (e.g., aging, neurodegeneration) may default to maladaptive mitophagy, contributing to bioenergetic failure.