Hypothesis

Aging neurons accumulate somatic copy-number variations (CNVs) that alter the surface expression of complement proteins and phagocytic ligands. These CNV‑induced changes act as a genomic‑burden ‘eat‑me’ tag, marking metabolically inefficient or weakly connected neurons for removal by microglia. In individuals with clonal hematopoiesis of indeterminate potential (CHIP), mutated monocytes infiltrate the brain and differentiate into microglia that exhibit heightened complement receptor (CR3) signaling and oxidative burst capacity, thereby amplifying the clearance of CNV‑laden neurons. This process represents an active quality‑control mechanism that optimizes neural network efficiency under declining energy budgets, rather than passive neurodegeneration.

Mechanistic rationale

1. CNV burden alters neuronal surfaceome – Large‑scale CNVs (>1 Mb) frequently encompass genes encoding complement regulators (e.g., CD46, CD55) or ligands for microglial receptors (e.g., CX3CL1, MFGE8). Dosage shifts from these CNVs change the balance of ‘don’t‑eat‑me’ versus ‘eat‑me’ signals, tilting neurons toward phagocytic recognition【https://elifesciences.org/articles/39217】.
2. CHIP‑derived microglia are primed for phagocytosis – CHIP monocytes carrying mutations in DNMT3A, TET2, or ASXL1 adopt a microglial phenotype with up‑regulated ITGAM (CR3) and CYBB (NOX2) expression, enhancing their ability to recognize and engulf opsonized targets【https://pmc.ncbi.nlm.nih.gov/articles/PMC10353941/】.
3. Energy‑budget coupling – Neurons with high CNV load exhibit mitochondrial dysfunction and reduced ATP production, lowering their capacity to sustain ‘don’t‑eat‑me’ signals (e.g., CD47). Simultaneously, aging astrocytes release less ATP, diminishing microglial inhibitory purinergic signaling, further tipping the milieu toward phagocytosis.
4. Feedback loop – Engulfment of CNV‑rich neurons releases intracellular nucleic acids that stimulate microglial TLR7/9, promoting a pro‑inflammatory state that recruits additional CHIP‑derived monocytes, reinforcing the clearance cycle.

Testable predictions

• Prediction 1: In post‑mortem human cortex, neuronal CNV burden (quantified by single‑cell DNA sequencing) will negatively correlate with surface levels of CD47 and positively correlate with iC3b deposition, irrespective of Alzheimer’s pathology.
• Prediction 2: Brains from CHIP carriers will show a higher proportion of microglia co‑expressing CR3 and phagocytic cups surrounding neurons with elevated CNV load compared to non‑CHIP age‑matched controls.
• Prediction 3: Experimental induction of neuronal CNVs (via CRISPR‑mediated segmental duplications) in mouse cortical cultures will increase microglial phagocytosis only when co‑cultured with CHIP‑mutant microglia, not with wild‑type microglia.
• Prediction 4: Pharmacological blockade of CR3 or complement C3 will rescue synaptic density and cognitive performance in aged CHIP mice without affecting overall microglial numbers.

Experimental approach

1. Human tissue: Perform single‑cell DNA and RNA sequencing on dorsolateral prefrontal cortex from donors stratified by CHIP status (peripheral VAF >2%) and age. Quantify neuronal CNV size/number, assess transcriptomic signatures of complement regulators, and perform immunostaining for iC3b and microglial activation markers (IBA1, CD68). Use spatial transcriptomics to map microglia‑neuron interactions.
2. Mouse models: Generate Vav1‑Cre; Dicer1^fl/fl chimeras to engraft CHIP‑mutant hematopoietic stem cells into wild‑type recipients. Introduce neuronal CNVs via in utero electroporation of plasmids bearing a 1 Mb duplication of chromosome 11 (containing Cd47). Assess phagocytosis using pHrodo‑labeled synaptosomes and in vivo two‑photon imaging of microglial processes.
3. Intervention: Treat aged CHIP mice with anti‑CR3 antibody or C3a receptor antagonist; evaluate behavioral outcomes (Morris water maze, novel object recognition) and synaptic protein levels (PSD‑95, synaptophysin).

Potential confounders and mitigations

• Neuroinflammation independent of CNV: Measure cytokine profiles to ensure observed phagocytosis correlates with CNV burden rather than global inflammation.
• Peripheral immune cell contribution: Use CCR2 knockout mice to exclude monocyte‑derived macrophages that are not microglial.
• Technical bias in CNV detection: Apply multiple single‑cell CNV callers and validate with bulk low‑pass whole‑genome sequencing.

Implications

If validated, this hypothesis reframes age‑related neuronal loss as a genetically guided, microglia‑mediated pruning process akin to developmental synaptic refinement. It suggests that enhancing, rather than suppressing, microglial phagocytic capacity—particularly in CHIP contexts—could preserve circuit efficiency by removing genomically compromised neurons before they become sources of pathological activity. Conversely, indiscriminate anti‑inflammatory therapies might impede this beneficial quality‑control mechanism, accelerating cognitive decline. Ultimately, targeting the CNV‑complement‑microglia axis offers a novel avenue to promote brain healthspan.