Finding. Cut ND4 and ND5 off complex I in the computer, and ND2 opens a hydrated P-side pose it barely visits when the pump is intact. Abettan, Lasham, Endres and Sharma run atomistic molecular dynamics on an intact enzyme and on a truncated model missing ND5 and ND4. The new ND2 conformational change on the positively charged side of the membrane is much cheaper in the truncated enzyme by free-energy calculations. Exposing ND2 to bilayer helps. The local hydration looks like a recent plant complex I:III2 cryo-EM. The authors treat that as a clue that ND2 might still be able to release protons, and they name residues for biochemistry to settle it.
Why this paper matters
The three antiporter-like subunits are the proton pumps. A growing structural story sends the pumped protons out through ND5. If that is exclusive, ND2 and ND4 are relays, not vents. A simulation that hydrates ND2 only when ND5/ND4 are gone is either an artifact of a wound or a hidden vent. Either way it is a hypothesis with a residue list, which is what a mitochondria desk wants from an MD paper.
The plant CI:CIII2 resemblance keeps it from being a purely in-silico hallucination. Still, resemblance is not a pumping number.
What they actually measured
Two models, atomistic MD, free-energy comparison, hydration, a plant-structure parallel, a mutagenesis agenda. No wet assay.
How to read the score
High 80s as a mechanistic complex I note. Confidence is medium because truncation plus simulation is a provocation, not a stoichiometry.
Caveats
Deleted subunits change everything around ND2. Lipid exposure is part of the trigger they acknowledge. Do not rewrite textbooks until the residues are mutated.
What to do with it
If you mutate ND2, start with their P-side list and score pumping, not only assembly. If you simulate intact CI, ask how rarely that hydrated ND2 pose appears with ND4/ND5 present. Pair it with the mycothiazole open-tunnel structure this week.