Finding. Metformin's gut story has a bacterial complex I. Isber, Rolain and Bittar show that the drug, at millimolar growth-stopping doses and at lower fermentation-stopping doses, hits bacteria that carry NDH-1, the NADH:quinone oxidoreductase that is mitochondrial complex I's cousin. Delete nuoF and E. coli keeps fermenting on metformin. Delete ndh (NDH-2) and it does not. Nicotinic acid plus magnesium, or pyruvate, put fermentation back, which points at NAD+ regeneration above pyruvate. Docking parks the drug in the NDH-1 quinone pocket. Staphylococcus aureus, which has no NDH-1, does not care. Lactobacillus, Bifidobacterium, and Akkermansia muciniphila appear to lack the homologue, which is a ready-made ecological prediction: those taxa should shrug at this particular metformin effect.
Why this paper matters
Host complex I versus AMPK versus gut flora has been a three-way fight. This preprint does not end it. It names a bacterial subunit, a bypass, and a phylogenetic split. If metformin enriches Akkermansia in people, one boring explanation is that it starves the NDH-1 crowd. That is a testable ecology, not a new slogan about "the microbiome."
What they actually measured
Fermentation, two NADH dehydrogenase mutants, metabolite rescues, docking, a short comparative-genomics list. No in vivo community edit in the abstract.
How to read the score
High 80s as a complex I-homologue mechanism. Confidence is high for the E. coli genetic split.
Caveats
Concentration. Docking. No mouse. Host mitochondria remain a live target. Do not stop metformin hepatic research.
What to do with it
If you profile stool on metformin, split taxa by NDH-1 presence. If you build nuo mutants in a gnotobiotic mouse, this is the drug-interaction paper. Pair it with this week's mammalian complex I structures: same fold, different kingdom, same quinone-site instinct.