In a cell lacking OPA1 (top), mitochondria (red and green) remain aloof, but the combination of long and short OPA1 isoforms spurs them to fuse (bottom).

Mitochondria malfunction unless they occasionally fuse with each other. Papers by Song et al. (page 749) and Griparic et al. (page 757) now elucidate the protein processing necessary for these mergers and pin down one of the responsible enzymes.

A protein matchmaker that promotes mitochondrial unions is OPA1, which is located in the organelle's inner membrane. The protein is faulty in dominant optic atrophy, an inherited form of blindness. OPA1's many varieties—there are eight mRNA splice variants, each of which encodes polypeptides that undergo further processing—fall into two categories: long and short. Yeast harbor a similar protein and require both lengths for fusion. But a previous study suggested that only the long form is responsible for mitochondrial fusion in mammals.

Song et al. wanted to determine what the short forms were doing. They engineered mouse fibroblasts that lacked the gene for OPA1 to manufacture different combinations of long and short variants. Long versions alone didn't spur mitochondria to get together, the researchers found. Nor did the short forms. But mitochondria coalesced in cells that produced a mixture of long and short OPA1. The researchers suspect that their results differ from the previous finding because that group used RNAi to knock down OPA1, and some residual short proteins might have persisted in the cells.

Several proteins might cut OPA1 down to size. Song et al. found that the enzyme Yme1L triggers cuts at one position in OPA1, suggesting that it directly or indirectly helps produce the short form. Work by Griparic et al. also implicated the enzyme in slicing OPA1 to promote fusion. Their results also suggest that another unidentified protein is involved. The next question for researchers to tackle is why fusion requires both types of OPA1.