A Role for the M9 Transport Signal of hnRNP A1 in mRNA Nuclear Export

Among the nuclear proteins associated with mRNAs before their export to the cytoplasm are the abundant heterogeneous nuclear (hn) RNPs. Several of these contain the M9 signal that, in the case of hnRNP A1, has been shown to be sufficient to signal both nuclear export and nuclear import in cultured somatic cells. Kinetic competition experiments are used here to demonstrate that M9-directed nuclear import in Xenopus oocytes is a saturable process. Saturating levels of M9 have, however, no effect on the import of either U snRNPs or proteins carrying a classical basic NLS. Previous work demonstrated the existence of nuclear export factors specific for particular classes of RNA. Injection of hnRNP A1 but not of a mutant protein lacking the M9 domain inhibited export of mRNA but not of other classes of RNA. This suggests that hnRNP A1 or other proteins containing an M9 domain play a role in mRNA export from the nucleus. However, the requirement for M9 function in mRNA export is not identical to that in hnRNP A1 protein transport.The transport of macromolecules between the nucleus and cytoplasm is a bi-directional process. The best understood aspect is the import of nuclear proteins that carry a basic nuclear localization signal (NLS)¹ like the simple NLS found in SV-40 T antigen or the bipartite NLS found in nucleoplasmin (Dingwall and Laskey, 1991). Proteins of this class are recognized by the heterodimeric importin receptor, composed of importin α and importin β (for review see Powers and Forbes, 1994; Melchior and Gerace, 1995; Görlich and Mattaj, 1996). The NLS binds directly to the importin α subunit. The importin NLS protein complex docks at the cytoplasmic face of the nuclear pore complex in an energy-independent manner (Newmeyer and Forbes, 1988; Richardson et al., 1988). Subsequently, the small GTPase Ran/TC4 (Melchior et al., 1993; Moore and Blobel, 1993) and a protein of unknown function named variously pp15, p10, or NTF2 (Moore and Blobel, 1994; Paschal and Gerace, 1995) are required for translocation of the NLS-containing complex through the nuclear pore complex.A second major class of imported macromolecules are the uracil rich small nuclear (U sn) RNPs. They do not have a basic NLS but instead have a bipartite nuclear targeting signal. This is composed of an essential signal formed when the Sm core proteins bind to the U snRNA and an additional signal, the trimethyl-guanosine (m₃G) cap, which depending on the cell type or the U snRNA is either essential or required for optimal U snRNP import efficiency (Fischer and Lührmann, 1990; Hamm et al., 1990; Fischer et al., 1993). Kinetic competition experiments have supported the conclusion that U snRNPs require different limiting factors than do NLS-containing proteins for their import and that U snRNPs do not bind to importin α (Fischer et al., 1991, 1993; Michaud and Goldfarb, 1991; van Zee et al., 1993). There is also preliminary evidence that additional different receptors may be required for the nuclear uptake of other RNA species (Michaud and Goldfarb, 1992).Similarly, RNA export from the nucleus relies on recognition of the RNA or RNP export substrates by saturable factors (Zasloff, 1983; Bataillé et al., 1990; Jarmolowski et al., 1994). As for import, evidence for the existence of RNA class-specific export receptors has been obtained from kinetic competition experiments (Jarmolowski et al., 1994). Two RNA-binding proteins have been directly shown to function in RNA export, a nuclear cap binding protein complex in the case of U snRNAs (Izaurralde et al., 1995a ) and the HIV-1 Rev protein in the case of RNAs containing a rev response element (Fischer et al., 1994, 1995). In the case of mRNAs, the best candidates for export mediators are the heterogeneous nuclear (hn) RNP proteins (for review see Piñol-Roma and Dreyfuss, 1993; Izaurralde and Mattaj, 1995).About 20 different hnRNP proteins have been characterized in vertebrate cells (for review see Dreyfuss et al., 1993). The association of hnRNP proteins with mRNA in the nucleus and the cytoplasm suggests that they may regulate and/or facilitate different aspects of gene expression. The possibility that hnRNP proteins might be directly involved in the nucleocytoplasmic trafficking of mRNA molecules was suggested by the observation that several hnRNP proteins, including A1, A2, D, E, I, and K shuttle continuously and rapidly between the nucleus and the cytoplasm and are associated with mRNA in both compartments (Piñol-Roma and Dreyfuss 1992, 1993; Michael et al., 1995a , b). Of these, the best studied example is hnRNP A1. An A1-like hnRNP protein has been shown by immunoelectron microscopy to be associated with a specific mRNA in transit to the cytoplasm through the nuclear pore complex in the insect Chironomus tentans (Visa et al., 1996a ). In mammalian cells, the amount of A1 which is in constant flux between nucleus and cytoplasm is striking. It has been estimated that at least 120,000 molecules of A1 are exported to the cytoplasm per minute but then rapidly reimported such that the steady state localization of A1 is nuclear (Michael et al., 1995a ). Taken together, these results suggest that A1 and other shuttling hnRNP proteins such as A2, D, E, I, and K could play a significant role in the transport of mRNA from the nucleus to the cytoplasm.One key in understanding how hnRNPs may facilitate mRNA export is to determine the signals that mediate their shuttling, i.e., their import into and exit from the nucleus. The nucleocytoplasmic transport of A1 has been recently studied in detail, and the signals that mediate shuttling have been identified (Michael et al., 1995b ; Siomi and Dreyfuss, 1995; Weighardt et al., 1995). Nuclear import of A1 is determined by a 38-amino acid sequence, termed M9, located near the COOH terminus of the protein between amino acids 268 and 305. Its fusion to cytoplasmic reporter proteins such as pyruvate kinase resulted in rapid import of the fusion protein into the nucleus (Siomi and Dreyfuss, 1995). However, the A1 NLS has no sequence similarity to classical protein NLSs such as that of SV-40 large T antigen or nucleoplasmin (Siomi and Dreyfuss, 1995).Surprisingly, M9 also acts as a nuclear export signal (NES). In heterokaryon shuttling assays this domain is necessary and sufficient to allow the export of heterologous proteins, such as the nucleoplasmin core domain (NPLc), which are normally retained in the nucleus (Michael et al., 1995b ). Thus, M9 alone can account for the shuttling of A1. Other hnRNPs such as A2 and B1 bear sequences with striking similarities to M9 (Siomi and Dreyfuss, 1995). Mutagenesis experiments indicate that the NES and NLS activities of M9 are either identical or overlapping as mutants which block M9 NLS activity also abolish NES activity (Michael et al., 1995b ). It is therefore possible that M9 is recognized in the nucleus and the cytoplasm by the same receptor.The second category of NES described was first found in the HIV-1 Rev protein and the inhibitor of protein kinase A (Fischer et al., 1995; Wen et al., 1995; Bogerd et al., 1996; for review see Gerace, 1995). These short, leucine-rich NES sequences bear no relationship to the primary sequence of M9. Furthermore, saturation of the export factor recognized by the Rev NES has no effect on mRNA export (Fischer et al., 1995). A model for mRNA export has been postulated on the basis of the hnRNP data described above. In this model, NES/NLS containing hnRNPs bind in the nucleus to mRNA molecules and deliver them, via the export pathway they access, to the cytoplasm. In the cytoplasm these hnRNP proteins dissociate from the mRNA and return to the nucleus. To further test this model we have analyzed the transport of hnRNP A1 and mRNA in Xenopus laevis oocytes. The oocyte offers a unique opportunity to manipulate specific import or export pathways, like that accessed by M9, and examine the effect on mRNA nuclear export. By using this approach we show here that M9 is, as in somatic cells, a functional NLS in oocytes. Moreover, competition studies indicate that M9 defines a novel class of NLS, since saturation of the M9- mediated import pathway does not interfere with the two previously identified import pathways used by classical NLS-bearing proteins or m₃G-capped-spliceosomal U snRNPs. Injection of an excess of hnRNP A1 but not of a mutant form of the protein lacking the M9 domain, resulted in a specific inhibition of mRNA export, demonstrating that the M9 domain is recognized by a saturable component of the mRNA export machinery. The export of other cellular RNAs such as U snRNAs and tRNA was, in contrast, not affected. Further analysis of mutant hnRNP A1 proteins provides evidence that M9 recognition during mRNA export differs from its recognition during protein transport.