Hypothesis

Aging‑associated loss of butyrate‑producing gut microbes reduces HDAC inhibition in hippocampal dentate gyrus neurons, removing an epigenetic brake on the complement cascade. Consequently, low‑activity, metabolically expensive neurons acquire C1q tags that flag them for microglial phagocytosis, turning synaptic pruning into outright neuronal eviction. This predicts that butyrate shortage shifts microglial surveillance from synapse‑specific remodeling to wholesale removal of inefficient cells, linking microbial metabolism directly to activity‑dependent quality control.

Mechanistic Model

1. Butyrate decline → ↓ histone H3K9/K14 acetylation in dentate gyrus (1).
2. HDAC hyperactivity → ↑ transcription of complement component C1q and downstream effectors in neurons with low firing rates (energy‑expensive but weakly connected).
3. Elevated neuronal C1q → binds to microglial CR3 receptors, priming phagocytic synapses and, when sustained, the soma itself (2).
4. Microglial shift → butyrate normally suppresses microglial activation (2); its loss removes this brake, converting surveillance from synaptic stripping to whole‑cell engulfment.
5. Feedback → evicted neurons reduce local glutamate release, further lowering network activity and reinforcing the pruning signal.

Predictions & Tests

• Prediction 1: In aged mice, hippocampal neurons showing low calcium activity will have higher C1q immunoreactivity than highly active peers; this difference will disappear after butyrate supplementation. Test: Combine in vivo calcium imaging with immunostaining for C1q; treat aged mice with sodium butyrate or butyrate‑producing probiotics and quantify C1q signal in low‑ vs high‑activity cells.
• Prediction 2: Transfer of old microbiota to germ‑free young mice will increase microglial C3 receptor expression and neuronal eviction, an effect blocked by HDAC inhibitors. Test: Perform microbiota transplant, measure microglial CD11b/CR3 levels, stereological neuron counts in dentate gyrus, and rescue with an HDACi such as suberoylanilide hydroxamic acid (SAHA).
• Prediction 3: Older adults with higher fecal butyrate will exhibit lower CSF C1q levels and better memory scores, independent of total neuronal density. Test: Correlate fecal butyrate (GC‑MS), CSF C1q (ELISA), and neuropsychological scores in a longitudinal cohort; adjust for age and total hippocampal volume via MRI.

If these experiments show that butyrate loss specifically elevates complement tagging on low‑activity neurons and that restoring butyrate rescues neuronal numbers without altering overall inflammation, the hypothesis gains support. Conversely, if butyrate manipulation changes global inflammation but not neuronal C1q patterns or survival, the idea of activity‑dependent eviction is refuted, pointing to a broader neuroprotective role instead.