Hypothesis

Elevated ketone bodies shift the microglial complement‑CD47 balance toward synaptic preservation by increasing neuronal CD47 "don’t eat me" signaling and suppressing microglial C1q "eat me" tagging, thereby raising the metabolic threshold for activity‑dependent synapse elimination during aging.

Mechanistic Rationale

1. Ketone bodies as signaling ligands – β‑hydroxybutyrate (βHB) inhibits class I histone deacetylases (HDACs) and activates PPARγ in neurons and microglia. HDAC inhibition promotes transcription of Cd47 (observed in cancer immunotherapy contexts) while PPARγ activation drives microglia toward an anti‑inflammatory, homeostatic state that reduces complement component C1q synthesis and release.
2. Impact on the pruning cascade – Higher neuronal CD47 raises the threshold for microglial phagocytosis of synapses tagged by C1q, shifting the efficiency‑based culling toward only the most severely hypoactive connections. Simultaneously, microglial PPARγ activation diminishes extracellular vesicle (EV) release that carries DNA‑damage signals known to stimulate proinflammatory phenotypes [6].
3. Integration with metabolic flexibility – When glucose is abundant and ketones low, HDAC activity rises and PPARγ signaling wanes, lowering CD47 and increasing C1q, thus permitting the age‑associated pruning of weakly active synapses described in the complement‑mediated model [1,2]. This creates a reversible, metabolism‑sensitive checkpoint that determines whether the brain’s "quality control" operates conservatively or aggressively.

Testable Predictions

• Prediction 1: In aged mice, a ketogenic diet (KD) or chronic βHB supplementation will increase neuronal CD47 immunoreactivity (measured by flow cytometry or immunohistochemistry) and decrease C1q deposition on synapses in the hippocampus and cortex compared with ad libitum fed controls.
• Prediction 2: The KD‑induced increase in CD47 will be abolished in neurons lacking Hdac3 (neuronal‑specific Hdac3 KO) or in microglia lacking Pparg (microglia‑specific Pparg KO), indicating that HDAC inhibition and PPARγ activation are necessary mediators.
• Prediction 3: Ex vivo hippocampal slices from aged mice treated with βHB will show reduced microglial engulfment of GFP‑labeled synapses (using pHrodo‑linked synaptophysin) and higher baseline synaptic density (via PSD‑95 puncta) relative to vehicle‑treated slices; this effect will be blocked by anti‑CD47 antibodies.
• Prediction 4: Behavioral readouts (Morris water maze, novel object recognition) will improve in KD‑fed aged mice, and this cognitive benefit will be lost when CD47 is blocked or when microglial PPARg is deleted.

Falsifiability

If ketogenic manipulation fails to alter neuronal CD47 levels, synaptic C1q tagging, or microglial phagocytic activity—as quantified by the assays above—then the proposed ketone‑mediated gating of complement‑CD47 balance is not supported. Likewise, if microglial PPARg deletion does not affect the KD’s protective effect on synapses, the hypothesized mechanistic link between ketone signaling and microglial phenotype would be refuted.

Experimental Outline (brief)

• Animals: 20‑month‑old C57BL/6J mice, plus conditional KO lines (Neuron‑Hdac3^fl/fl; Cx3cr1‑CreER for microglia‑Pparg^fl/fl).
• Interventions: KD (70% fat, 10% carb) or βHB (1 g/kg i.p. daily) for 8 weeks; controls receive standard chow or saline.
• Readouts: Western blot/IF for CD47 and C1q; ELISA for soluble C1q in CSF; flow cytometry for microglial activation markers (CD68, CD86, Arg1); synaptophysin/PSD‑95 immunostaining; microglial synaptophagy assay using pHrodo‑synaptophysin beads; cognitive testing.
• Statistical plan: Two‑way ANOVA (diet × genotype) with post‑hoc Tukey; n≥10 per group to detect 20% effect size with 80% power.

By directly linking a circulating metabolite to the molecular “eat‑me/don’t‑eat‑me” axis that governs synaptic pruning, this hypothesis reframes age‑related synapse loss not as inexorable decay but as a metabolically tunable quality‑control process—offering a novel therapeutic lever to preserve cognitive function in aging.