Ironing out the kinks: splicing and translation in bacteria.

Abstract

In eukaryotic cells, RNA processing is physically separate from protein synthesis. As nuclear export of unspliced RNA is restricted, only mature messages are normally exposed to the translational machinery. In bacteria, however, splicing must be coordinated with the translation of nascent transcripts. These two processes make very different demands on the RNA substrate: Splicing of autocatalytic introns requires that the 58 and 38 splice sites be brought together as part of an elaborate tertiary structure, whereas translation requires that the mRNA be relatively free of secondary structure. Nonetheless, introns have been found in highly expressed genes in eubacteria, bacteriophages, mitochondria, and chloroplasts (for review, see Burke 1988). Clearly, there must be some means of balancing splicing of bacterial introns with cotranscriptional translation. In this issue, Semrad and Schroeder (1998) provide the surprising answer that splicing of a group I intron from phage T4 is facilitated by translation of the upstream open reading frame (ORF). This enhancement of splicing is achieved by modulating the long-range conformation of the premRNA. Their results provide useful analogies for the coupling of eukaryotic pre-mRNA splicing with transcription. Bacterial mRNAs exclusively contain group I or group II introns, and the three group I introns that are present in phage T4 are all able to self-splice in vitro (for review, see Belfort 1990). The introns are found in genes encoding thymidylate synthase (td), ribonucleotide reductase (nrdB) (Belfort 1990), and anaerobic ribonucleotide reductase (sunY, or nrdD) (Young et al. 1994). In addition to sequences that provide the necessary functions of selfsplicing, td and sunY in trans contain internal ORFs that encode double-stranded DNA endonucleases (Belfort 1989). The endonucleases trigger homing, or site-specific movement of the intron sequences to intronless alleles. Self-splicing requires that the intron RNA fold into a unique secondary and tertiary structure (Cech and Herschlag 1996). The central core of this structure is highly conserved among group I introns and contains the active site where the transfer of phosphodiester bonds takes place. A helix containing the 58 splice site docks into the active site via hydrogen bonds with its ribose 28 hydroxyl groups (Pyle et al. 1992). Recognition of the 38 splice involves several weak interactions, including a 2-bp stem between the 38 end of the intron and nucleotides in the intron core (Burke et al. 1990). The folded structure of the RNA depends on coordination of magnesium ions, which are required for self-splicing (Cech and Herschlag 1996). Long-range interactions, such as base-pairing between hairpin loops, or tetraloop–helix receptor interactions, also stabilize the tertiary structure of the catalytic core (for review, see Brion and Westhof 1997). Recent experiments on the folding kinetics of group I introns, as well as experiments carried out in the 1970s on tRNA, have begun to tease out the mechanisms by which RNAs reach a biologically active conformation (for review, see Draper 1996; Brion and Westhof 1997). Small hairpins can form in 10–100 μsec, and tRNAs can fold within milliseconds (Draper 1996). In contrast, larger RNAs fold in stages over much longer periods of time. The tertiary structure of the P4–P6 domain of the Tetrahymena group I intron, which can fold independently, appears in a few seconds (Sclavi et al. 1997). The core of the intron, however, takes several minutes to become completely folded (Zarrinkar and Williamson 1994; Banerjee and Turner 1995). The very slow folding of longer RNAs arises, in part, from their tendency to form many alternative secondary structures. As RNA secondary structure is stable, incorrect base pairs have the potential to trap the molecule in inactive conformations that can persist for relatively long periods of time (for review, see Herschlag 1995; Thirumalai and Woodson 1996; Brion and Westhof 1997). These metastable states may be quite dissimilar to the final structure. Therefore, the folding process itself can result in RNA populations with different levels of biological activity. Misfolding of RNA has been shown to inhibit ribozyme activity and spliceosome assembly (Goguel et al. 1993; Zavanelli et al. 1994; Uhlenbeck 1995). As discussed below, competition between metastable RNA conformations can also serve as a normal mechanism by which gene activity is regulated. Estimates of in vivo splicing rates are 10to 50-fold faster than in vitro self-splicing (Brehm and Cech 1983; Zhang et al. 1995). This disjunction between in vitro and in vivo activity of catalytic RNAs implies that kinetic folding traps are normally overcome in the cell. Proteins E-MAIL sw74@umail.umd.edu; FAX (301) 314-9121.