key: cord-0038959-sd9fuid9
authors: King, Andrew M.Q.
title: RNA viruses do it
date: 2002-12-19
journal: Trends Genet
DOI: 10.1016/0168-9525(87)90173-9
sha: b92577975d4be41796efb347f92d626ffc5084c0
doc_id: 38959
cord_uid: sd9fuid9

nan

nitor normal protein. Possible alternative derivations for the amyloid peptide would include aberrant splicing or a mutation, perhaps somatic, which predisposes the protein to cleavage.

An even more exciting finding is that the gene encoding the amyloid protein is located on chromosome 21, a finding that has been confirmed by Robakis et al.17 . Obviously this could be chance, but it is very tempting to speculate that overexpression of the amyloid protein caused by the extra chromosome 21 in Down's syndrome is responsible for the symptoms of Alzheimer's disease shown by these patients. Although the modest increase to 1.5 times the normal gene do~age might not be expected to La~,e s-,ch a profound effect, it is not chffic~Jt to imagine that the primary dosage could result in a non-linear amplification in gene expression. This couldbe tegted by measuring mRNA and protein levels in the brains of aged Down's syndrome patients. Another potcn-tially revealing experiment would be to look for genetic linkage between the locus encoding the amyloid protein and the Alzheimer's disease locus segregating in families. If the two loci are the same, isolation of the gene encoding the amyloid protein will have been a major landmark in the study of human disease. Homologous recombination has been observed in every DNA genome in which it has been sought, but the extent to which RNA genomes recombine is still a matter for conjecture. It was not until 1982 that recombination in RNA was first demonstrated conclusively by fingerprinting viral genomes produced by crossing different strains of foot-andmouth disease virus 1. Subsequently, another member of the picornavirus family, poliovirus, was also shown to recombine 2 and, more recently, so also was mouse hepatitis virus a, a coronavirus. Both these groups of animal viruses have genomes consisting of a single positive-sense RNA molecule. Since replication takes place in the cytoplasm, without a DNA intermediate, the only way these viruses could exchange genetic information was by a novel kind of sequence rearrangement between RNA molecules.

The newest member of the RNA sex club, brome mosaic virus, Is quite different from the others. Several variants of this tripartite plant virus were found to have exchanged 3' terminal sequences between genome segments 4. The recombining sequences were similar to each other, and some of the crossovers took place at homologous sites. The discovery of a third RNA virus group that recombines homologously has encouraged speculation that perhaps all RNA viruses do it.

The importance of RNA recombination is not merely academic. These days, most cases of poliomyelitis in Western countries are caused by neurovirulent revertants of the attenuated vaccine virus; according to Kew 5 , reversion is frequently associated with genetic recombination between the three poliovirus serotypes of which the vaccine is usually composed.

Recombination, as exemplified by picurnaviruses, has three main characteristics.

(1) It is homologous. To date, the sequences of almost 50 genetic cross-overs have been reported, by several groups. These show that recombination occurs without introducing insertions, deletions or base substitutions, even though the parental sequences in the region of the cross-over may not match each other perfectly e.

(2) Recombination is a general, r,~ther than a site-specific process. When crossing closely related virus strains, we found many different cross-over sites scattered throughout the genome 7. Not surprisingly, the mor~ distantly related the parental strains, the more recombination is biased towards conserved regions of the genome. However, even when different serotypes are crossed, the number of potential recombination sites is very large; in one study e. nine sites were distinguished within a sequence of just 190 nucleotides. We should not be misled by the various models that have appeared in the literature recently (one is particularly ingeniousS), which invoke specific folding of the RNA template at the site of the cross-over. None of these studies of possible base-pairing inchded any controls and, whatever role RNA folding may play in recombination, there is no evidence of specificity in either primary or secondary structure at cross-over sites.

(3) Recombination is intrinsically very efficient. This property of TIG --March 1987, VoL 3, no. 3 picornaviruses is not generally apparent because, in the quest for sequence information, nearly all recent studies have been devoted to crosses between different serotypes. However, in the genetic mapping studies of isogenic mutants, done many years ago °'m, the proportion of recombinants in the virus yield typically approached 1%. This may not sound impressive, until one considers that the assay only detects cross-overs that take place (a) between the loci of the selectable markers, (b) in one direction and (c) between different parents. Allowing for these factors, we can estimate that 10-20% of viral genomes undergo recombination during each cycle of infection. Recent data e confirm this estimate. The high incidence of multiple cross-overs 7 in a single growth cycle is also consistent with a high recombination frequency.

To answer this question, Kirkegaard and Baltimore e recently performed the following elegant experiment. A poliovirus mutantj carrying separate mutations for temperature sensitivity and guanidine resistance, was crossed with wild-type virus. These genetic markers enabled the growth of each parent to be inhibited independently during mixed infections. When the wild-type parent was allowed to replicate at the restrictive temperature in the presence of a non-replicating pool of mutant genomes, the two viruses recombined normally. However, when parental roles were reversed using guanidine selection, recombination was completely inhibited.

The implication that one parent plays an active mating role, while the other is passive, strongly suggests a copy-choice mechanism for RNA recombination (Fig, 1) . Since it was the active partner that contributed the 3' end of its genome to the offspring, it follows that the viral replicase switches templates while synthesizing negative-sense P,.NA. In addition, the high frequency of homologous cross-overs implies some mechanism for specifying the correct reinitiation site on the new template; this is most simply explained by base-pairing with the 3' end of the growing RNA strand.

This model is attractive, because several RNA virus transcriptases are known to generate sequence re- Modified from ReL 6.

arrangements as part of their normal function, by mechanisms that resemble copy-choice recombination. In coronaviv, tses, for example, all snbgenornic mRNAs have a common 5' terminal leader sequence derived from another region of the genome. Lai and his co-workers have suggested that genetic recombination in coronaviruses is a manifestation of the same process 3. Why RNA viruses recombine is unclear. Different virus strains may benefit from the increased genetic diversity that results from exchanging information. Indeed, there is evidence that some of the major advances in the evolution of RNA viruses were r~inational events II. However, this is not necessarily the reason that picernaviruses recombine. The one demonstrable advantage of recombination is that it eliminates deleterious mutations and, in view of the high mutation rates of RNA genomes, this seems a more plausible function.

Modern Approaches to Vaccines

As this issue of TIG was going to press we heard of the tra.q;c death of Steve Prentis in a car accident. Steve launched TIG in 1985 and edited the journal during its first year of publication. He will be sadly missed by his many friends and colleagues throughout the world,