The RNA-dependent RNA polymerase of bacteriophage Qb was first isolated in 1965 by Haruna and Spiegelman, who named it Qb replicase (1). Today, this enzyme still represents the prototype of RNA virus-replicating polymerases, and it has remained unique in that it allows the specific replication and amplification of an infectious viral RNA by a complex of soluble, stable, and defined protein components in vitro. Early studies had established that the viral RNA genome (the plus strand) was replicated by primer-independent end-to-end RNA synthesis in the 5′ to 3′ direction. A free, single-stranded complementary strand (minus strand) is synthesized as an intermediary product and serves as a template for the synthesis of single-stranded plus strands, whereas double-stranded RNA is devoid of any template activity (2). Investigations with Qb replicase have resulted over the years in a surprising number of novel concepts of more general relevance, such as: (1) specific template recognition by a polymerase (1); (2) recruitment of host proteins as subunits of a viral polymerase (3, 4); (3) role of a viral polymerase in the temporal control of protein biosynthesis (2); (4) an internal RNA binding site as an enhancer of synthesis initiation (5); (5) activation of an RNA template by a host factor acting as an RNA molecular chaperone (6); (6) secondary and tertiary structure as determinants of template recognition (7, 8); (7) role of secondary structure in elongation of RNA synthesis (9, 10); (8) sequence evolution in vitro (11) and the quasispecies concept (12); (9) de novo synthesis of replicatable RNA (13); (10) site-directed mutagenesis (14); (11) use of small RNA templates as vectors for foreign sequences (15); and (12) development of such systems for diagnostic purposes (16). Most of these aspects will be discussed here.
1. Protein components of the in vitro Qb RNA replication system
Qb replicase is isolated from Qb-infected Escherichia coli cells as an enzyme consisting of four different subunits, only one of which (subunit b, 65.5 kDa) is phage-specific. The other subunits are recruited from the host cell and were identified as the ribosomal protein S1 (subunit a, 61.2 kDa) and the protein synthesis elongation factors EF-Tu and EF-Ts (subunits g, 43.2 kDa, and d, 30.3 kDa). In addition to this holoenzyme form of replicase, a smaller core enzyme species, consisting of only subunits b, g, and d, was found to be equally active for all templates except the Qb plus strand. Of the three core subunits, b (phage-specific) represents the catalytic subunit, and EF-Tu/Ts (subunits g and d) appear to be important for the assembly and stabilization of the active enzyme complex. The role of protein S1 (subunit a) as an RNA-binding protein consists of mediating the recognition of the Qb plus-strand RNA as a template (see below). The template activity of the Qb plus strand, in contrast to the other known template RNAs, depends further on the presence of an additional host-derived RNA binding protein called the "host factor" (17), or Hfq protein (6x11kDa). Hfq protein was recently found to function in the cell as an "RNA chaperone" required to activate translation for the induction of the stationary phase and other functions (18).
Replicases of other RNA phages make use of the same host-derived subunits (19, 20), but their host factor requirements may be different, as shown for phage f2 (21) and its close relative MS2 (22).
2. Template Specificity
For an RNA to be replicated and amplified by replicase, both the RNA itself and its complementary strand must be efficient templates. This is the case for Qb RNA and the Qb minus strand in the presence of the holoenzyme and host factor. The core enzyme without the host factor is sufficient to copy the minus-strand RNA into a plus strand, but cannot by itself use the latter as an efficient template. However, the core enzyme is efficient for replicating and amplifying a family of small RNAs, collectively called 6 S RNA, that arise in vitro (and possibly also in vivo) in the absence of added template RNA, either by uninstructed de novo synthesis or through elongation and recombination of small RNA fragment contaminants (see below). RNAs of closely related phages (eg, SP) have partial activity (23), but more distant phage RNAs (eg, MS2), like other viral and cellular RNAs, are inactive. Synthetic templates like poly(C) and mixed polyribonucleotides rich in C function as templates for the synthesis of a complementary strand, which however remains paired with the template and is itself not active as a template.
3. Recognition of the Qb plus-strand template
Replicase holoenzyme forms tight complexes with the Qb plus strand by binding at two internal RNA regions called the S- and M-sites (24), mapping at positions 1247 to 1346 and 2545 to 2867 of the Qb genome (4217 nucleotides), respectively. Electron microscopic examination of such complexes revealed the formation of RNA loops between these sites, showing that the two interactions occurred simultaneously on one and the same strand (25). Similar looped complexes were observed by binding of S1 protein alone, which suggests that these internal interactions are mediated by the a subunit (26). Binding of replicase to the S-site, located at the start of the coat protein cistron, is not necessary for template recognition but is thought to represent a translational repression mechanism by which replicase prevents the attachment of ribosomes during the time it uses the plus strand as a template (2). This stands in contrast to the M-site, in which a branched secondary structure element (nucleotides 2696 to 2754) has an enhancer-like function essential for high template activity (5). Binding of the 3′-end, ie, the site of initiation of synthesis, is mediated by the host factor, which in addition binds to two internal sites adjacent to the M- and S-sites (26, 27). Qb phages adapted to growth in host factor-less E. coli strains contained mutations altering the 3′-terminal folding of the RNA and releasing the sequence at the immediate 3′-end from base-pairing (6). This is in agreement with the concept of the host factor as an RNA chaperone that modifies the secondary and tertiary structures of the 3′-end so as to make it accessible to replicase.
In summary, the recognition of the Qb plus strand by replicase occurs through contacts at three (or possibly even more) different RNA sites that map far apart on the genome but must be held in close vicinity by the tertiary structure of the RNA (8). Interestingly, it is the host proteins S1 and Hfq that are the mediators of these interactions, whereas the catalytic b subunit is very probably responsible for the direct contacts at the 3′-end of the template that allow synthesis to start. Initiation occurs with GTP, which is complementary to the template’s essential 3′-terminal H CCC motif; the additional A residue found at the 3′-end of most RNA chains is not essential.
4. The Qb minus strand and other RNAs as templates
The interactions of replicase with the Qb minus strand and the replicating 6 S RNAs are expected to be different from those with the plus strand, because the former RNAs are efficient templates for core replicase in the absence of proteins S1 and Hfq. Deletion analysis of the Qb minus strand identified two structural features required for template activity (7). One consists of a sequence that folds into an imperfectly base-paired stem-loop structure located near the 5′-terminus of the minus strand. A highly homologous sequence is found in MDV-1 RNA (a 6 S RNA) and was characterized as an element essential for recognition of this template (28). The other essential element is formed by two short complementary sequences forming a helical stem by long-range base-pairing near the 3′-end of the minus strand. On the basis of stability calculations, the 3′-terminal H CCC(A) sequence, which represents another essential structure, appears in the minus strand to exist in an unpaired conformation, in agreement with the fact that this template does not require a host factor.
Of the many short-chain RNAs characterized as templates, their only common features (29) consisted of a single stranded H CCC(A) 3′-end, as well as sequences rich in secondary structure, as discussed below.
Recently, the SELEX technique was used to generate two families of short RNA ligands that bound strongly to replicase holoenzyme and were active templates (30). One contained a pseudoknot structure rich in A and C residues in the unpaired loops and was found to crosslink to S1 protein. The other contained a polypyrimidine tract and crosslinked to EF-Tu (the g subunit). At present, it appears difficult to integrate these findings into a common picture with those from the Qb plus and minus strand.
5. Secondary structure required for the release of single-stranded product RNA
Short-range intrastrand base-pairing in product and template RNAs appears to be necessary for the release of the product RNA in a single-stranded form, presumably by competitively inhibiting the formation of long product-template duplexes. In fact, longer tracts of template sequence unable to form local stem-loop structures resulted in the formation of duplexes between product and template strands that could not be replicated further (9). Moreover, RNA phage mutants carrying insertions of such tracts in vivo quickly evolved to revertants that had eliminated the inserts (10).
6. Evolution in vitro and in vivo, error rates, recombination, and the question of de novo synthesis
Spiegelman and coworkers recognized very early that their RNA amplification system provided a unique opportunity to study evolution in vitro by serial transfer experiments (11). Studies focusing on this aspect were instrumental in the development of the "quasi-species" concept by M. Eigen (12). Estimates of the phage Qb mutation rate (about 10-4; (31)) and of the sequence heterogeneity of the Qb genome in phage populations determined by Weissmann and coworkers (32) were in full agreement with this notion.
The occurrence of RNA recombination in RNA phage replication was first experimentally demonstrated in vivo (33), but had been suspected before, because sequences homologous to cellular RNAs were found in several 6 S RNA species (34). Later in vitro studies confirmed the process, which was thought to take place via copy-choice (35) or by a novel type of mechanism (36).
The claim that rigorously purified replicase under specific conditions can be observed to synthesize small replicating RNAs de novo without template instruction (13) encountered opposition because of the difficulty of excluding contamination by traces of 6 S RNA in any laboratory engaged in in vitro work with replicase (37). The doubts have stimulated, however, a large amount of careful work, making today a convincing case for de novo synthesis, such as the demonstration that different incubation mixtures of identical ingredients result in entirely unrelated RNA sequences arising after widely fluctuating latency periods (29). The stochastically arising RNA species and their propagation and evolution to high-efficiency replication could be followed quantitatively by fluorescence techniques in a sophisticated capillary incubation apparatus (13).
7. Practical applications of replicase
Not surprisingly, the unique properties of replicase as a nucleic acid amplification system have raised interest in developing practical uses. Most efforts were directed toward using replicating RNAs as vectors for the amplification of foreign sequences (15). Success was achieved mostly with relatively short inserts, because of the high tendency of the system toward mutation and recombination and the requirement of a strong secondary structure, as outlined above. Nevertheless, a sensitive diagnostic assay using short inserts of ribosomal RNA sequences from several pathogenic agents was developed and marketed as an alternative to PCR (16).
8. Outlook
Despite the described attempts to get "something useful" out of Qb replicase, it appears likely that the main fascination with this enzyme will remain within the realm of basic science. The most intriguing questions may concern the precise mechanics of the interplay of the macromolecular components that make the system behave in the way it does. No doubt, knowledge of the three-dimensional structures of these components would be a great step forward toward this understanding. Efforts toward determining a structure of replicase by X-ray crystallography have not been successful thus far, but should be continued and extended to substructures like the core enzyme, S1 protein, and host factor. As the crystallography of RNA is still in its infancy, large RNA structures like the Qb plus and minus strand may not be accessible very soon, but interesting information could come from specific and well-studied small RNAs like MDV-1, especially if co-crystals with replicase and its subassemblies could be analyzed. In the meantime, the recent rapid advances in computer modeling of RNA structures (38) will continue and, together with phylogenetic comparisons and chemical or biochemical probing techniques, should give us improved models of template RNAs and their complexes with replicase. Similarly, we can expect that functional studies with site-directed RNA mutants, and especially the elegant evolutionary approach in which deficient mutants are allowed to evolve back to high viability (8), will continue to advance our understanding of this exemplary RNA-protein recognition process.
