Finding the answer wasn't hard (I got lucky). But the answer only leads to more questions.
It turns out very few organisms manufacture the "big" manganese catalase. The organisms in question belong to just 3 genera: Rhizobium, Bradyrhizobium, and Rhodopseudomonas. (The closely related Mesorhizobium has a Mn catalase, but it's the "normal" size Mn catalase. Meanwhile, its cousin, Sinorhizobium, has no Mn catalase.)
The first 280 or so amino acids of the Mn catalases made by Rhizobium, Bradyrhizobium, and Rhodopseudomonas align quite well with the normal-size Mn catalases made by other organisms, the only difference being the 140-amino-acid trailer on the end of the "long versions." I used the alignment editor in Mega6 (tantamount to Notepad) to Cut the trailer portion out of one of the sequences and Paste it into the BLAST search field at UniProt.org. A search against the protein database revealed something quite interesting: The trailer portion of the Bradyrhizobium Mn catalase is a 45%-identity match for the enteric-bacteria yciF gene (which is quite a good match considering the phylogenetic distance between E. coli and Bradyrhizobium).
|The E. coli yciF gene (top) is a 58% match for the Bradyrhizobium yciF gene (middle), which in turn is a 64% match for the trailer portion of katN (manganese catalase gene) of Bradyrhizobium, bottom.|
Further investigation left little doubt that the "big" Mn catalases are indeed fusion proteins: yciF fused to the C-terminal end of the Mn catalase.
But: What in the world is yciF? It turns out to be a very widely distributed, highly conserved protein of uncertain function. The protein has been purified and its crystal structure determined at 2.0 Å resolution, but its function is still uncertain. What we do know is that it is stress-inducible (along with other yci-series genes) and seems to bind a metal ligand, probably iron; and it shares structural features with rubrerythrin, a non-heme iron protein implicated in oxidative stress protection in anaerobic bacteria and archaea. In E. coli strains that have Mn catalase, the yciF gene occurs two genes upstream (on the 5' side) of the catalase (katN) gene.
Because yciF is more widely distributed (and more highly conserved) than manganese catalase, and because most Mn catalase producers (including those with the fusion enzyme) have an additional, separate copy of yciF, it seems likely (to me, anyway) that the fusion protein was created by chance in the common ancestor of the Rhizobiales when the original katN gene was laid down by a phage or other mobile genetic element. (In E. coli, katN often occurs near phage genes.) Sinorhizobium lost the combo gene entirely, while Mesorhizobium (which makes a small Mn catalase) either obtained katN on its own or lost the 3' trailer from its fusion protein over time.
Since Bradyrhizobium (and the others) already have a separate yciF gene, it's a mystery why the trailer portion of the fusion gene continues to exist. It might very well provide a favorable enhancement of katN function somehow (maybe exploiting iron in an auxiliary catalytic center). If the trailer's doing nothing useful, it should have disappeared over time. (Maybe it did disappear from other Rhizobiales members, and just hasn't disappeared yet in the three genera that still make the "big" enzyme.) I have a feeling the trailer piece does do something useful. Unfortunately, no one has characterized the Braydrhizobium Mn catalase yet, experimentally. We'll probably have to wait until that happens to find out what the "big" enzyme is capable of.