Hypothesis: Intermittent fasting reprograms the brain’s complement‑mediated pruning machinery by simultaneously lowering neuronal C1q expression through SIRT1 activation and boosting astrocytic secretion of the complement inhibitor CD59. This dual action raises the threshold for synapse tagging, preserving weakly active but metabolically adaptable connections while still allowing elimination of irreversibly damaged ones.

Mechanistic rationale: Fasting elevates β‑hydroxybutyrate, which functions as an endogenous HDAC inhibitor and increases NAD+ levels. The resulting SIRT1 activation deacetylates NF‑κB p65, reducing its transcriptional drive on the C1q gene in neurons and astrocytes [1][2]. Concurrently, fasting stimulates TFEB‑mediated lysosomal biogenesis in astrocytes, enhancing their capacity to load and secrete CD59‑containing exosomes [3]. Elevated extracellular CD59 binds to C8 and C9, blocking membrane attack complex formation and thereby shielding synapses from microglial phagocytosis even when C1q is present [4].

Testable predictions:

1. In young adult mice undergoing alternate‑day fasting (ADF), neuronal C1q mRNA in the hippocampus will decrease by ~30% relative to ad libitum controls, an effect abolished by SIRT1 inhibition (EX‑527) [5].
2. Astrocyte‑specific CD59 knockout will prevent fasting‑induced rescue of spine density despite normal ketone levels, as measured by two‑photon imaging of dendritic spines after 4 weeks of ADF.
3. CSF exosomes isolated from fasted humans will show higher CD59:C1q ratios and lower synaptosomal C1q deposition in vitro compared with fed controls; this ratio will correlate negatively with plasma β‑hydroxybutyrate and positively with performance on a pattern separation task.
4. Pharmacological blockade of C3 (using Cp40) will not further improve synaptic preservation in fasted SIRT1‑intact mice, indicating that fasting acts upstream of C3 activation.

Falsifiable outcomes: If ADF fails to reduce neuronal C1q transcription regardless of SIRT1 activity, or if astrocytic CD59 loss does not diminish the protective effect of fasting on synapses, the hypothesis is refuted. Similarly, if exosomal CD59:C1q ratios do not predict cognitive outcomes in fasting humans, the proposed peripheral biomarker would be invalid.

This framework integrates metabolic signaling, complement regulation, and glial‑neuronal communication to explain how the brain can remain energetically efficient without sacrificing functional plasticity. It shifts the view of age‑related synaptic loss from a passive degenerative process to a maladaptive shift in the complement‑pruning set point that can be corrected by modulating metabolic flexibility.