Finding. In human Alzheimer synapses, mitochondrial proteins are already down by the middle of the Braak march, while the proteostasis and astrocyte coats are still climbing. Georgiades, Jackson and colleagues isolate synaptoneurosomes from 67 middle-temporal gyri (40 Alzheimer cases across stages, 27 controls) and pull six stage-linked proteomic signatures. Mitochondrial function and synaptic signalling fall by mid-stage. Clathrin-mediated endocytosis rises then stalls. Proteostasis and perisynaptic astrocytic proteins rise the whole way. Excitatory and inhibitory protein signals drift (P = 7.80 times 10 to the minus 3). They then take those brain signatures into two plasma proteomic sets and nominate 42 proteins. PHGDH is the one that survives early CDR 0-to-0.5, neuropathological progression, and the excitatory-inhibitory imbalance.
Why this paper matters
Synapse loss is still the best tissue correlate of Alzheimer cognition. Timing the mitochondrial drop to mid-stage, ahead of a still-rising proteostasis/astrocyte program, is a staging fact you can put next to a mouse. Plasma PHGDH is a hypothesis, not a test. The n and the Braak span are why the mitochondrial clause is believable.
What they actually measured
Human synaptoneurosome proteomics, networks, pathway/cell-type scores, plasma mapping. No synaptic respirometry.
How to read the score
Mid 80s. Human, staged, mitochondrial signature named. Confidence is high for the tissue pattern.
Caveats
Post-mortem. Protein is not flux. One gyrus. Do not sell PHGDH as a diagnostic.
What to do with it
If you stage Alzheimer synapses, put mitochondria and signalling in the mid-stage bin and proteostasis/astrocyte on the progressive axis. Pull PHGDH only as a plasma echo of that map. Pair it with this week's ARMS2 and TBI-pyruvate papers: different diseases, same instinct that carbon and organelles fail before the late anatomy.