The oligomerization and ligand-binding properties of Sm-like archaeal proteins (SmAPs)

Excision of noncoding regions (introns) is a vital step in the maturation of precursor mRNAs. Most eukaryotic protein-coding genes contain multiple introns (Long et al. 1995), and thus high-fidelity pre-mRNA processing is essential to ensure production of mature mRNAs with correctly registered exons. The simultaneous excision of introns and splicing of exons in eukaryotic pre-mRNA is catalyzed by a transiently stable assembly of five small nuclear ribonucleoproteins (snRNPs). This large assembly of uridine-rich snRNPs (U snRNPs) is known as the spliceosome, and at various stages in its catalytic cycle it consists of the U1, U2, U4/U6, and U5 snRNPs (Yu et al. 1999). Five small nuclear RNAs (snRNAs) and at least 80 proteins are contained within the spliceosome (Burge et al. 1999), making it roughly the same size as the ribosome (sedimentation coefficient of ~60S; Muller et al. 1998); furthermore, assembly of U snRNPs into spliceosomes is likely to be independent of pre-mRNA binding, as suggested by recent isolation of a novel U1•U2•U4/U6•U5 penta-snRNP devoid of mRNA (Stevens et al. 2002). Extensive biochemical and genetic data have shown that a key step in snRNP assembly is stepwise binding of seven cytoplasmic Sm proteins to exported snRNAs (Will and Luhrmann 2001). Each U snRNP is a complex formed from an ~110–180-nucleotide (nt) snRNA and two classes of proteins: (1) snRNP-specific proteins that confer snRNP-specific functions (e.g., U1A protein of U1 snRNPs) and (2) the Sm or Sm-like (Lsm) proteins that are common to each snRNP core (Will and Luhrmann 1997). The snRNAs contain a single Sm or Lsm binding site with the uridine-rich consensus sequence PuAU~4–6GPu (Pu = purine). However, specificity for this sequence is not stringent and there can be redundancy in Sm-snRNA binding (Jones and Guthrie 1990). The Sm sites are predicted to be single-stranded RNA regions flanked by stem-loop structures (Burge et al. 1999; Yu et al. 1999). Sm binding is highly sensitive to modifications of the flanking stem-loops and the Sm site of a given snRNA, and varies from one snRNA to another (Jarmolowski and Mattaj 1993). Sm-snRNA binding also may be modulated by interactions between certain Sm proteins and the survival of motor neurons (SMN) protein complex (Selenko et al. 2001), and by symmetric dimethylation of arginine residues in some of the RG dipeptide repeats of Sm (Brahms et al. 2000; Friesen et al. 2001; Meister et al. 2001) and Lsm (Brahms et al. 2001) proteins by a putative “methylosome” (Friesen et al. 2002). In eukaryotes, Sm D1•D2 and E•F•G heteromers simultaneously bind to snRNA to yield a “subcore” snRNP complex (Raker et al. 1996, 1999; Will and Luhrmann 2001). The final component to join the Sm complex is the B/B′•D3 heterodimer, and this triggers hypermethylation of the 5′ m7G cap of snRNA to a trimethylated guanosine cap (m3G). The m3G cap and the snRNA•Sm core complex form a bipartite nuclear localization signal that results in transit of the snRNP core to the nucleus, where association of various snRNP-specific proteins completes the assembly process. The importance of Sm proteins in RNP assemblies is underscored by their phylogenetic distribution: In addition to the canonical Sm and Lsm proteins found in eukaryotes ranging from yeast to humans, an Sm-like archaeal protein (“SmAP”) family has been discovered (Salgado-Garrido et al. 1999; Mura et al. 2001). The recent demonstration that the E. coli bacteriophage host factor Hfq is an Sm-like protein provides the first example of a eubacterial Sm protein (Moller et al. 2002; Zhang et al. 2002). These results imply fundamental roles for Sm proteins in the early evolution of RNA metabolism. Sm proteins probably mediate critical RNA-RNA, RNA-protein, and protein-protein interactions in snRNP cores. The vast network of protein-protein interactions in which Sm proteins participate was recently suggested by genome-wide two-hybrid screens of yeast Lsm proteins (Fromont-Racine et al. 2000). Sm proteins have a tendency to associate into cyclic oligomers. Prompted by biochemical and genetic data, electron microscopic (EM) investigations of U snRNP particles revealed the “doughnut-shaped” ultrastructure of Sm and Lsm cores (Kastner et al. 1990; Achsel et al. 1999). The realization that Sm and Lsm proteins occur in groups of at least seven paralogs within the genome of a given organism suggests that snRNP cores are formed from Sm heteroheptamers, and two recent results verify this. First, Stark et al. (2001) reconstructed a 10 Å-resolution map of the U1 snRNP by cryo-EM and found that a model of the Sm heptamer could be docked into the ring-shaped body of the snRNP. Next, the in vivo stoichiometry of Sm proteins in yeast spliceosomal snRNPs was determined by a differential tag/pull-down assay, showing that the snRNP core domain contains a single copy of each of the seven Sm proteins (Walke et al. 2001). Stable subheptameric Sm complexes have been suggested as intermediates along the snRNP core assembly pathway (e.g., a D1•D2•E•F•G complex that binds snRNA; Raker et al. 1996), and ultracentrifugation and EM show that some of these oligomers can form ring-like structures that resemble intact, heptameric snRNP cores (e.g., a (E•F•G)2 hexamer in Plessel et al. 1997). Such findings emphasize the importance of cyclic Sm heptamers in the snRNP core, and raise the possibility of other oligomeric states. There is no atomic-resolution structure of a eukaryotic snRNP core. Nonetheless, the crystal structures of Sm-like archaeal proteins from Afu (Toro et al. 2001), Pae (Mura et al. 2001), and Mth (Collins et al. 2001) reveal a cyclic Sm homoheptamer and provide a model for snRNA binding in the snRNP core. Sm monomers fold as strongly bent, five-stranded antiparallel β-sheets (Kambach et al. 1999a) and form toroidal heptamers that surround a conserved cationic pore. The inner surface of this pore appears to be the oligouridine-binding site. The similarity between SmAP1 monomer and dimer structures and the nearly identical human Sm D1•D2 and D3•B heterodimers (Kambach et al. 1999b) supports SmAP-based models for the heptameric snRNP core.