Finding. Diplonema papillatum does not keep a cabinet of pentatricopeptide-repeat DYW (PPR-DYW) deaminases, one per mitochondrial RNA site. It keeps one homolog, PPRD1. Valach, Aguilar, Burger and colleagues show that PPRD1 works with a near-stoichiometric PolX-like partner, DAPX1. Knock either protein down and the cell slows, and both C-to-U and A-to-I editing collapse across five mitochondrial clusters that together hold 110 sites.
Why this paper matters
Mitochondrial genomes are sloppy. Many lineages fix the mess after transcription, by deaminating cytidines (and sometimes adenosines) so the messenger RNA can be translated. Land plants industrialized that fix: tens to hundreds of PPR-DYW proteins, each typically aimed at one site. The authors say that wherever this family was retained, it expanded independently, except in diplonemids.
That exception is the story. A single deaminase covering more than 100 mitochondrial sites is either a promiscuous wrecking enzyme or a core catalytic unit that borrows specificity. The biochemistry argues for the second reading.
The two-protein core
Native pull-down of PPRD1 recovered about 20 proteins, most of them sub-stoichiometric, including helical-repeat proteins that look like RNA binders. One partner did not look like a visitor. DAPX1 came down with PPRD1 in roughly equal amounts, and the interaction went both ways. Structural modelling puts DAPX1 against the deaminase domain, as a stabilizer and as extra surface for other factors to grab.
The functional test is blunt. Silence PPRD1 or DAPX1 and growth drops. In vivo, C-to-U editing falls, and so does A-to-I editing, across the same five mitochondrial clusters. A plant-style one-protein-one-site model does not predict that dual-base, multi-cluster collapse from two knockdowns.
How to read it
The working model is a core (PPRD1 plus DAPX1) plus disposable specificity factors (the sub-stoichiometric set). That is how you get accuracy without 100 dedicated enzymes, and how you keep off-target deamination from shredding the transcriptome. It is also why A-to-I shows up here at all: the same core is required, whether or not the DYW domain itself is the adenosine deaminase.
Score 80. This is not a clinical paper. It is a clean hit on organelle gene expression: a lineage that refused the usual PPR-DYW expansion and still recodes mitochondrial RNA at industrial scale. Confidence is high for the in-vivo dependence and the pull-down stoichiometry, lower for the modelled DAPX1 geometry.
Caveats
No reconstituted enzyme is claimed in the abstract. PPRD1 could still be a required scaffold for a different catalytic subunit. DAPX1's structural job is a model. The 20 partners are a list, not a site map. Diplonemid editing will not automatically explain plant, slime-mold, or kinetoplastid systems.
What to do with it
If you track mitochondrial RNA processing, this is the diplonemid counterexample to PPR family expansion. If you build editors, the design hint is a shared deaminase core plus exchangeable RNA-binding adapters. Pull the partner table and the 110-site editing matrix before you write “PPR-DYW equals site-specific C-to-U.”