Hypothesis

In aging entorhinal cortex, hyperphosphorylated tau does not kill neurons indiscriminately; instead, it serves as a molecular tag that flags synapses with low activity for microglial removal. Persistent synaptic loss drives homeostatic downscaling of the host neuron, reducing its metabolic load. When synaptic pruning crosses a critical threshold, the neuron itself becomes a target for complement‑mediated phagocytosis, effectively evicting the cell as an energy‑saving measure. This links the seed idea—neuronal loss as an efficiency‑driven process—to the observed selective vulnerability of excitatory grid cells in tauopathy.

Mechanistic Reasoning

1. Activity‑dependent caspase‑3 activation marks inactive synapses for microglial phagocytosis [3].
2. Tau accumulation preferentially occurs at synapses with low firing rates, stabilizing microtubules in a way that further suppresses vesicle release (see excitatory neuron loss in EC‑tau models) [2].
3. Chronic low activity leads to sustained caspase‑3 signaling, amplifying synaptosis.
4. As synapse number falls below a set point, the neuron upregulates complement C1q and C3 tags, signaling microglia to engulf the soma—a pathway known from developmental synaptic pruning and disease models.
5. The result is a selective loss of metabolically expensive, weakly connected excitatory grid cells, while inhibitory interneurons, which maintain higher baseline activity, are spared [1].

Testable Predictions

• Prediction 1: In EC‑tau mice, pharmacological enhancement of neuronal activity (e.g., chemogenetic activation of layer II excitatory cells) will reduce tau accumulation at synapses and delay both synapse loss and neuronal death.
• Prediction 2: Blocking microglial phagocytosis (e.g., with annexin V or anti‑C3 antibodies) will preserve synapse numbers despite high tau, but will not prevent somatic tau buildup.
• Prediction 3: Conversely, silencing excitatory grid‑cell activity in wild‑type mice will increase synaptic caspase‑3 activation and promote tau‑like phosphorylation at inactive synapses, mimicking the early tauopathy signature.

Falsifiability

If activating excitatory neurons fails to lower synaptic tau or protect synapses, or if inhibiting microglia does not rescue synapse loss despite reduced activity, the hypothesis that tau tags inactive synapses for activity‑dependent eviction would be refuted. Likewise, if silencing activity does not increase synaptic caspase‑3 or tau‑like modifications, the proposed causal link between inactivity and tau tagging would be unsupported.

By framing tau as an activity‑dependent signal that bridges synaptic pruning to whole‑cell removal, this hypothesis rewrites the narrative from passive toxic damage to an adaptive—though maladaptive in aging—efficiency drive.