by this enzyme has in fact been demonstrated for brome mosaic virus (BMV), a plant virus with a 3'-terminal tRNA-like structure very similar to that of Qß (Rao et al., 1989).
These observations about the Qß genome led us to propose that the first tRNA-like structures arose as ''genomic tags" that marked the 3' ends of ancient RNA genomes for replication by RNA enzymes in the RNA world (Weiner and Maizels, 1987; Maizels and Weiner, 1993). The simplest such tags would have been the predecessors of the "top half" of modern tRNA, consisting of a coaxial stack of the TΨC arm on the acceptor stem (see Figure 5). The presence of such 3' terminal tRNA-like structures in two different kingdoms—contemporary bacterial viruses like Qß and plant viruses such as turnip yellow mosaic virus (TYMV) and BMV (reviewed by Hall, 1979; Haenni et al., 1982; Guerrier-Takada et al., 1988)—suggests that these structures date back at least as far as the progenote. The role of tRNA in replication appears to have arisen much earlier, however. As we discuss in greater detail below, molecular fossil evidence suggests that tRNA-like structures were first used for replication of RNA genomes by RNA enzymes in the RNA world and then persisted through the transition from RNA to DNA genomes.
An Early Origin for tRNA Rationalizes the 5' and 3' Processing of tRNAs in Modern Cells. The genomic tag model immediately explains the existence of two enzymes with otherwise puzzling activities in contemporary tRNA processing, RNase P and tRNA nucleotidyltransferase. Transcription of tRNA typically initiates not at the 5' end of the functional molecule, but at a site upstream. The 5' leader sequences are then removed from the "pre-tRNA" by an endonucleolytic cleavage catalyzed by RNase P. RNase P is a ribonucleoprotein, but the RNA component alone is capable of catalysis (Guerrier-Takada et al., 1983). The presence of a catalytic RNA component suggests that RNase P activity arose in the RNA world (Alberts, 1986); use of an RNA component cannot be explained by the need to recognize so many different species of tRNA, since elongation factor Tu accomplishes the same task without the help of RNA. Further attesting to the ancient origin of RNase P, the structure it recognizes is highly conserved: the Escherichia coli enzyme can cleave the 3' tRNA-like structure of TYMV (Green et al., 1988; Guerrier-Takada et al., 1988; Mans et al. , 1990). But why would a tRNA processing enzyme be present in the RNA world? We suggest that the first function of RNase P may have been to free catalytic RNAs from the 3'-terminal genomic tags required for replication; cleavage may have been necessary to activate catalytic function or to prevent replication from interfering with catalysis. If this is the case, it would explain the puzzling fact that tRNAs undergo 5' processing at