Mito.newsMito.news
← All articlesEditorial brief · abstract-levelScore 72/100Confidence high
biorxiv2026-09-08methodsneurobiologyOXPHOSsingle-cell

A widely used TBI atlas called microglia neurons, then invented a KCNC3 OXPHOS signature

Most cells labeled “glutamatergic neuron” in CEREBRI, the dominant single-cell traumatic-brain-injury atlas, are glia. That contamination manufactured a biphasic ion-channel trajectory and a KCNC3-specific oxidative-phosphorylation signature that vanish when the labels are audited. Mitochondrial stories in public atlases are only as good as the cell-type names.

Mito.news · at a glance

Signal profile (abstract-level)

methods · neurobiology · OXPHOS · single-cell

Score 72/100BIORXIVhigh confidencemethods
72
Importance
42
Mito signal
53
Dysfunction
75
Evidence
30
Translational

Editorial signal profile from the abstract (importance score, mito keywords, dysfunction tags, evidence density, translational cues). Not a figure reproduced from the preprint PDF.

Finding. The most-cited single-cell traumatic-brain-injury atlas, CEREBRI, labeled 45,051 cells as glutamatergic neurons. Zhang, Gao and colleagues say 963 of those cells (2.1%) earn the name. The rest are microglia, astrocytes, oligodendrocytes, and leftovers. That lie was tidy enough to invent biology: a biphasic trip for 40 ion-channel genes, a KCNC3-specific oxidative-phosphorylation signature, and a day-7 inversion of KCNC3. Clean the labels and KCNC3 simply goes up early and stays up, in four datasets and three injury models. A trajectory fingerprint sends 96.3% of the informative ion-channel paths back to non-neuronal cells.

Why this paper matters for mitochondria

Plenty of TBI papers now talk about neuronal OXPHOS rewiring. If the neurons were glia, the OXPHOS module was a glial module wearing a stolen badge. AnnoAudit is the protocol: four checks (markers, clustering, margin-gated modules, applicability-gated models) rolled into an Annotation Contamination Score, plus a fingerprint that names the guilty cell type. CEREBRI scores 82.6%. A human ALS atlas scores 70.4%. Marker-defined controls pass. You only need the public count matrix and a canonical panel.

How to read the score

Low seventies. Methods, not a new respiratory-chain subunit. High confidence that the CEREBRI glutamatergic set is contaminated, and that a flagship mitochondrial-looking signature was downstream of that contamination. If your corpus only wants wet OXPHOS, this is a warning label. If your corpus trains agents on GEO, it is required reading.

Caveats

ACS is only as honest as the marker panel and the simulation. Corrected KCNC3 is still transcriptomics. The original atlas will keep circulating until people stop pointing at GSE269748 as ground truth.

What to do with it

Quarantine CEREBRI “glutamatergic neuron” OXPHOS and KCNC3-OXPHOS claims. Re-annotate before you build TBI mitochondrial networks. If you ship atlas-mining agents, add an ACS-like gate. The biology to keep is sustained acute KCNC3 up-regulation in real neurons, not a chimeric OXPHOS story.

Free HTML is above. Bots pay for JSON at /api/v1/papers/10-64898-2026-09-03-749125. Optional wallet tester: MetaMask ($0.005).

Source preprint

AnnoAudit: a marker-based protocol for auditing single-cell atlas annotations reveals annotation-driven artifacts in a widely used traumatic brain injury resource

10.64898/2026.09.03.749125

Zhang L, Rao H, Li M, Qian X, Zhang Y, Yan Q, Gao R.

Related briefs