Extensive nonhomologous recombinations occur between the 5' and 3' fragments of a replicable RNA in a cell-free system composed of pure Qbeta phage replicase and ribonucleoside triphosphates, providing direct evidence for the ability of RNAs to recombine without DNA intermediates and in the absence of host cell proteins. The recombination events are revealed by the molecular colony technique that allows single RNA molecules to be cloned in vitro. The observed nonhomologous recombinations are entirely dependent on the 3' hydroxyl group of the 5' fragment, and are due to a splicing-like reaction in which RNA secondary structure guides the attack of this 3' hydroxyl on phosphoester bonds within the 3' fragment.

The ability of RNAs to spontaneously rearrange their sequences under physiological conditions is demonstrated using the molecular colony technique, which allows single RNA molecules to be detected provided that they are amplifiable by the replicase of bacteriophage QL L. The rearrangements are Mg 2+ -dependent, sequence-non-specific, and occur both in trans and in cis at a rate of 10 39 h 31 per site. The results suggest that the mechanism of spontaneous RNA rearrangements differs from the transesterification reactions earlier observed in the presence of QL L replicase, and have a number of biologically important implications.z 1999 Federation of European Biochemical Societies.

Q␤ replicase (RNA-directed RNA polymerase of bacteriophage Q␤) exponentially amplifies certain RNAs (RQ RNAs) in vitro. Here we characterize template properties of the 5 and 3 fragments obtained by cleaving one of such RNAs at an internal site. We unexpectedly found that, besides the 3 fragment, Q␤ replicase can copy the 5 fragment and a number of its variants, although they lack the initiator region of RQ RNA. This copying can occur as a 3-terminal elongation or through de novo initiation. In contradistinction to RQ RNA and its 3 fragment, initiation on these templates occurs without regard to the 3-terminal or internal oligo(C) clusters, is GTP-independent, and does not result in a stable replicative complex capable of elongation in the presence of aurintricarboxylic acid. The results suggest that, although Q␤ replicase can initiate and elongate on a variety of RNAs, only some of them are recognized as legitimate templates. GTP-dependent initiation on a legitimate template drives the enzyme to a "closed" conformation that may be important for keeping the template and the complementary nascent strand unannealed, without which the exponential replication is impossible. Triggering the GTP-dependent conformational transition at the initiation step could serve as a discriminative feature of legitimate templates providing for the high template specificity of Q␤ replicase.Q␤ replicase, the RNA-directed RNA polymerase of bacteriophage Q␤, amplifies the 4217-nt 1 -long genomic Q␤ RNA and a number of RQ RNAs, which are usually Յ250 nt in length. The natural source of RQ RNAs is Q␤ phage itself or Q␤ phageinfected Escherichia coli cells where these RNAs are formed by recombination from viral and/or cellular RNAs and propagated (1-3). Recently, many new RQ RNAs have been selected from random (4) or artificially designed (5) sequences, or produced by in vitro RNA recombination (6, 7). The amplification of these RNAs is exponential as long as the enzyme is in molar excess: the number of RNA molecules doubles in each round of replication, because both the original RNA and its complementary copy are replicase templates. Approximately 10 4 copies of a single genomic RNA molecule are produced in a Q␤ phageinfected E. coli cell in less than 1 h (8). Amplification of small RQ RNAs is much faster: up to 10 10 copies are produced at room temperature within 10 min in a cell-free system comprised of purified Q␤ replicase and all four NTPs (3), and this is the absolute record of the rate of nucleic acid amplification. The high amplification rate allows single RQ RNA molecules to rapidly produce detectable molecular colonies if they are amplified in a gel (9, 10).However, Q␤ replicase does not amplify most RNAs, including any tested cellular RNAs or genomic RNAs of other viruses. Selection experiments indicated that only a few of the initial diversity of 10 12 unique sequences of 50 -77 nt in length are replicable (4), demonstrating a very high degree of template specificity of the enzyme. Now, almost 40 years since its discovery (11),...

An earlier developed purified cell-free system was used to explore the potential of two RNA-directed RNA polymerases (RdRps), Q␤ phage replicase and the poliovirus 3Dpol protein, to promote RNA recombination through a primer extension mechanism. The substrates of recombination were fragments of complementary strands of a Q␤ phage-derived RNA, such that if aligned at complementary 3-termini and extended using one another as a template, they would produce replicable molecules detectable as RNA colonies grown in a Q␤ replicase-containing agarose. The results show that while 3Dpol efficiently extends the aligned fragments to produce the expected homologous recombinant sequences, only nonhomologous recombinants are generated by Q␤ replicase at a much lower yield and through a mechanism not involving the extension of RNA primers. It follows that the mechanisms of RNA recombination by poliovirus and Q␤ RdRps are quite different. The data favor an RNA transesterification reaction catalyzed by a conformation acquired by Q␤ replicase during RNA synthesis and provide a likely explanation for the very low frequency of homologous recombination in Q␤ phage.Recombinations (sequence exchange and rearrangements) between and within RNA molecules are rare but biologically important events contributing to the evolution and diversity of RNA viruses (1, 2) and generating defective interfering RNAs that attenuate viral infections (3). In contrast to splicing and other types of regular RNA rearrangements, recombinations occur without apparent sequence or structure specificity (1, 2). There are indications that recombination may occur between cellular RNAs (4 -6), eventually resulting, by means of reverse transcription and integration, in alterations in the chromosomal DNA. Spontaneous Mg 2ϩ -catalyzed rearrangements in RNA sequences (7) might have been a mechanism for evolution in the prebiotic RNA world and might have evolved into contemporary sequencespecific ribozyme-catalyzed reactions (8, 9).RNA recombination was discovered more than 40 years ago as an exchange of genetic markers between polioviruses (10, 11), and since then similar approaches were used to demonstrate that genomes of RNA viruses of animals, plants, and bacteria are all capable of recombination (2, 4, 12, 13). However, such in vivo experiments utilizing living cells, as well as in vitro studies that used crude cell lysates could not uncover the underlying molecular mechanisms or even definitely answer the question if recombining entities were RNA molecules themselves or their cDNA copies, because every living cell contained enzymes capable of reverse transcription and appropriate dNTP substrates. It became evident that further progress in this field depended on the availability of adequate in vitro systems whose composition and other parameters can be strictly controlled by the experimenter (2, 14).The first example of such a sort has been the cell-free system employing purified Q␤ replicase, RNA-directed RNA polymerase (RdRp) 1 of bacteriophage Q␤ (15). The system...

Transcription of bacteriophage N4 middle genes is carried out by a phage-coded, heterodimeric RNA polymerase (N4 RNAPII), which belongs to the family of T7-like RNA polymerases. In contrast to phage T7-RNAP, N4 RNAPII displays no activity on double-stranded templates and low activity on single-stranded templates. In vivo, at least one additional N4-coded protein (p17) is required for N4 middle transcription. We show that N4 ORF2 encodes p17 (gp2). Characterization of purified gp2 revealed that it is a single-stranded DNA-binding protein that activates N4 RNAPII transcription on single-stranded DNA templates through specific interaction with N4 RNAPII. On the basis of the properties of the proteins involved in N4 RNAPII transcription and of middle promoters, we propose a model for N4 RNAPII promoter recognition, in which gp2 plays two roles, stabilization of a single-stranded region at the promoter and recruitment of N4 RNAPII through gp2-N4 RNAPII interactions. Furthermore, we discuss our results in the context of transcription initiation by mitochondrial RNA polymerases.