By backcrossing both APPSwe/PS1dE9 and Csf1r ΔFIRE/ΔFIRE lines onto CBA (95-97%), the authors avoided the hydrocephalus/lethality that plagued the prior 5xFAD x FIRE KO cross . Flow cytometry, IBA1 IHC and microglia-enriched gene RNA-seq confirmed near-total microglia absence, and transgene expression plus the 99:83 APP CTF ratio were matched between APP-WT and APP-FIRE, a key control .
Plaque side: plaque coverage and number fell across all three probes (6E10, MX-04, OC; p<0.0001, n=6-8 mice/genotype) and SDS-insoluble Aβ dropped, yet the plaques that did form had similar dense-core volume (p=0.706) but a smaller peri-core fibrillar (OC+) halo and increased OC solidity (p=0.030, p=0.033). A decision tree using plaque morphology predicted microglial status with 92% accuracy (73 test plaques, 14 animals), independently corroborating the structural difference .
Downstream side: reactive astrogliosis (GFAP+ fraction of ALDH1L1+ astrocytes), synapse loss (array tomography, SYP/PSD-95) and SMI312+ neurite dystrophy all persisted in plaque-containing fields without microglia—although each was quantitatively attenuated, likely reflecting the thinner fibrillar Aβ halo. Whole-cortex RNA-seq is the starkest result: mutant APP induced 813 protein-coding DEGs in APP-WT but only 22 in APP-FIRE (p_adj<0.05) .
The finding is genuinely counterintuitive: it contradicts the classical clearance view of microglia and partially contradicts Grathwohl's earlier conclusion that plaques form normally without microglia . The authors note prior CSF1R-inhibitor work was confounded by incomplete killing and off-targets , whereas constitutive FIRE deletion gives a clean, complete genotype. Proposed mechanisms—endolysosomal Aβ concentration favoring seeding and TAM-receptor-mediated core formation —are plausible but untested here. The persistence of synapse loss without microglia (and without C1q) challenges the complement-phagocytosis model , pointing instead to neuron-intrinsic excitotoxicity or astrocyte glutamate dyshomeostasis.
No behavior/cognition, no tau pathology (APP/PS1 lacks tangles), no acute slice physiology, and no test of whether timing of microglia loss matters—so the "early-stage therapy" conclusion is inference, not demonstration. Replication in an independent microglia-null system would be decisive.
| Outcome | APP-FIRE vs APP-WT | Statistic | n (mice) |
|---|---|---|---|
| Plaque coverage/number (3 probes) | Reduced | p<0.0001 | 6 vs 8 |
| MX-04 dense core volume | No change | p=0.706 | 7 vs 7 |
| OC fibrillar halo volume | Reduced | p=0.030 | 7 vs 7 |
| Reactive astrocytes (plaque-proximal) | Present, attenuated at 25-100 µm | p=0.0003 | 5 vs 7 |
| Synapse loss (plaque fields) | Persists without microglia | p<0.0001 | 6 vs 8 |
| Neurite dystrophy density | Persists, reduced | p=0.0024 | 7 vs 7 |
| Cortical DEGs | 813 → 22 | p_adj<0.05 | 5-6/group |
| Vascular Aβ (Collagen IV colocalization) | No change | ns | 5 vs 7 |
Know what changed, what holds up, and what remains uncertain. Every Friday. No ads.