Separate information required for nuclear and subnuclear localization: additional complexity in localizing an enzyme shared by mitochondria and nuclei.

The TRM1 gene of Saccharomyces cerevisiae codes for a tRNA modification enzyme, N2,N2-dimethylguanosine-specific tRNA methyltransferase (m2(2)Gtase), shared by mitochondria and nuclei. Immunofluorescent staining at the nuclear periphery demonstrates that m2(2)Gtase localizes at or near the nuclear membrane. In determining sequences necessary for targeting the enzyme to nuclei and mitochondria, we found that information required to deliver the enzyme to the nucleus is not sufficient for its correct subnuclear localization. We also determined that mislocalizing the enzyme from the nucleus to the cytoplasm does not destroy its biological function. This change in location was caused by altering a sequence similar to other known nuclear targeting signals (KKSKKKRC), suggesting that shared enzymes are likely to use the same import pathway as proteins that localize only to the nucleus. As with other well-characterized mitochondrial proteins, the mitochondrial import of the shared methyltransferase depends on amino-terminal amino acids, and removal of the first 48 amino acids prevents its import into mitochondria. While this truncated protein is still imported into nuclei, the immunofluorescent staining is uniform throughout rather than at the nuclear periphery, a staining pattern identical to that described for a fusion protein consisting of the first 213 amino acids of m2(2)Gtase in frame with beta-galactosidase. As both of these proteins together contain the entire m2(2)Gtase coding region, the information necessary for association with the nuclear periphery must be more complex than the short linear sequence necessary for nuclear localization.