Introduction

Viruses are a class of genetic element, dependent on suitable host cells for their propagation. Thus, the evolutionary development of viruses must have been contingent on the prior evolution of potential host cells, and the genetic material of viruses must primordially have derived from cellular nucleic acids. Many virus coded proteins, particularly enzymes, have characteristic sequence similarities to non-viral proteins, and this is taken to indicate that many viral genes have cellular genes as progenitors. Having made these almost axiomatic assertions, it is by no means easy to fill in further strategic detail on virus origins. We would like to know what kind of pre-viral elements gave rise to early viruses; what forces acted in the genesis and evolution of viruses; at what epochs in the evolution of cellular organisms viruses appeared; and to what degree contemporary viruses have radiated from common progenitors. This chapter aims to examine some aspects of these questions, but it will quickly become apparent just how limited is our capability in obtaining firm or detailed answers.

Two general classes of data on viruses are applicable to considering details of viral origins, evolution and relatedness. The first includes what we know of basic phenotypes in terms of virion structure (including gross structure of the genome) and replicative cycle; ‘higher’ aspects of phenotype, namely properties relating to pathogenesis and disease, are less useful for these purposes. The second class consists of direct information on genotypes, that is, determined DNA sequences and their interpretations in terms of gene organization and encoded protein sequences.

Introduction

The third variable domain (V3) of the external envelope glycoprotein (gpl20) of the human immunodeficiency virus type 1 (HIV-1) contains determinants of virus cytopathicity, cell tropism and virus infectivity (Cheng-Mayer, Shioda & Levy, 1991; Chesebro et al, 1991; Freed & Risser, 1991; Hwang et al, 1991; Ivanoff et al, 1991; Takeuchi et al., 1991; de Jong et al, 1992; Fouchier et al, 1992). Antibodies binding to the virus V3 region block cell-free virus infectivity and cell-to-cell spread of virus (Goudsmit et al., 1988; Palker et al, 1988; Rusche et al, 1988). Recent evidence suggests that the ability to induce V3-specific antibodies may be pivotal to vaccine efficacy. Berman et al (1990) and Girard et al (1991) have demonstrated that chimpanzees can be protected by active immunization from infection by virions. Subsequently, Girard and colleagues have shown that the vaccination protocol used previously (Girard et al, 1991) protected against challenge with infected cells (M. Girard, personal communication). Both Berman and Girard suggested that antibodies to the virus V3 region contributed to protection, a notion that was corroborated by the fact that monoclonal antibodies to the V3 domain protected chimpanzees against infection by HIV-1 virions (Emini et al, 1992). Variation of the HIV-1 envelope protein and of the V3 domain in particular, may control HIV-1 pathogenesis and vaccine efficacy by modulating virus virulence and sensitivity to neutralizing antibodies.

Two major forms of variation are observed during the course of infection: one general and one V3-specific. Host-to-host transmission has a major impact on one form of variation, but hardly one on the other, resulting in very important principles of V3 diversification in the population.

Introduction

Although the sequences of the various genome segments of members of the eight genera that constitute the Reoviridae family have diverged completely, they still possess several common functional motifs, and the structural proteins that they encode have retained the ability to form structurally similar virus particles. The evolution of reoviruses has proceeded via genetic drift driven by positive and negative selection (acquisition of new hosts and ability of proteins to accommodate amino acid substitutions, respectively), which included not only point mutations but also partial genetic duplications, and proceeded in its early stages, at least under some conditions, via transitions rather than transversions. Evolution also proceeded via genetic shift which, for these viruses (rigid icosahedral particles) was successful only when the newly introduced genome segments resulted in the formation of virus particles no less stable than those generated by the parental homologous genome segment set.

Members of the Reoviridae family possess unique genomes that consist of 10, 11 or 12 segments of dsRNA that vary in size from 600 to 4000 bp. They form eight genera the genome segments of whose members have diverged to the point of complete randomness, do not re-assort (that is, cannot be incorporated into each others’ genomes), and encode proteins which although retaining key functions, possess no common epitopes. The question of the evolution of reoviruses can be examined both within these eight genera, and among them.

Introduction

The tobamovirus group of plant viruses has 12 recognized members (Francki et al., 1991) sharing the following characteristics: infective particles are very stable, rigid rods, of about 300 × 18 nm, with a sedimentation coefficient of 190 S; each particle is built of about 2000 protein subunits of a single molecular species (Mr 17–18 × 103), helically (pitch 2.3 nm) arranged together with a monopartite, single-stranded, genomic RNA (Mr 2 × 106, or ∼ 6.5 kb) of messenger polarity.

The genomic RNA (gRNA) encodes at least four proteins: gRNA is the mRNA for a 126 kD protein (126 K), whose open reading frame (ORF) initiates 60-71 nt from the capped 5′ end, and for a readthrough protein of 183 kD (183 K). Both the 126 K and the 183 K are required for viral RNA replication, and found in them are consensus sequences for methyltransferases and helicases in the 126 K and for RNA-polymerases, in the 54 kD (54 K) readthrough portion of the 183 K. A third ORF encodes a 30 kD protein (MP), required for the cell-to-cell movement of the virus in the plant; the fourth ORF encodes the coat protein (CP). To the 3′ side of the CP ORF, there is a non-coding region (3′ ncr) 179-414 nt long, that may adopt a tRNA-like configuration and can be amino-acylated, for most tobamoviruses, with histidine. Both MP, CP (and perhaps the 54 K readthrough part of the 183 K), are translated from 3′ -coterminal subgenomic RNAs produced during replication.

Introduction

Analyses of populations of viruses with RNA genomes find a large amount of variability that results from re-assortment of genome segments of those viruses that have multipartite genomes, recombination events, point mutations and small deletions or insertions. The reverse transcriptases (RT) and the RNA-dependent RNA polymerases (RdRp) cause the last three types of change. These polymerases are error-prone owing to their intrinsic low level of fidelity and to their lack of correction mechanisms. In addition, host factors can also be responsible for certain mutations.

This chapter deals with point mutations; indeed, point mutations represent the main source of diversity occurring during replication of the genome of RNA viruses. It attempts to briefly survey 1) the genetic variation in virus populations, 2) the methods used to determine mutation rates and frequencies, 3) the estimates of such mutation rates and frequencies obtained in vivo and of error rates obtained in vitro for the most studied viruses, and 4) our present knowledge concerning the mechanisms of point mutations. It does not deal with recombination and re-assortment; the reader is referred to other chapters of this book for information on these topics.

Genetic variation in virus populations

Most RNA virus infections correspond to heterogeneous virus populations that can be referred to as quasi-species; this term describes complex distributions of replicating molecules subject to mutation and competitive selection (Domingo et al., 1992). If RNA virus populations are maintained at constant selective pressures, genomic variability is high, but a stable distribution is reached that usually undergoes minimal changes (Domingo et al., 1978).

The idea to collate and edit a book on virus evolution occurred to each of us quite independently, but when we discovered that we had this common desire, it seemed reasonable to pool our resources, experiences, and energies. We assumed that, whereas none of us had a firm grasp of either the questions or the answers, collectively we might be able to attract as participants many people who have given virus evolution more than passing thoughts. Further, we did not want to edit a book containing ‘great thoughts by great people’. From the outset, we wanted to bring together current data-based views of the evolution of various viruses. Considering that thousands of viruses have been recognized, it was an obvious impossibility to try to cover the entire field. Indeed, we have made a concerted effort not to do so. Keeping in mind Duncan McGeoch's dictum that ‘Viruses may be looked on as mistletoe on the Tree of Life’, we fully expected that evolutionary practice would vary, and vary considerably from one family to another.

Therefore, we thought it appropriate to convene a meeting of people we expected would provide a basis for organizing a book on the subject and who might provide at least a glimmer of a consensus. We organized a symposium ‘Co-evolution of viruses, their hosts and vectors’ in Madrid from December 9–11, 1991, and invited some of the colleagues we knew to be working directly or indirectly on virus evolution, and advertisements attracted others who wished to join us.

Sequence mutation is the most common mechanism for the evolution of RNA viruses. However, major changes of viral genotype often involve exchanges of RNA segments between viruses, as exemplified by influenza virus evolution. For viruses with non-segmented RNA genomes, genetic exchange was previously thought to be rare. In recent years, however, sequence analyses of viral RNA revealed that many RNA viruses have apparently evolved from other viruses by exchange or rearrangement of genome sequences. Furthermore, experimental systems have been established for several viruses, in which recombination could be demonstrated between two viral RNAs. Thus, genetic recombination appears to be more important in RNA virus evolution than was previously realized.

Mechanistically, RNA recombination is similar to the generation of defective-interfering (DI) RNA, as both involve polymerase jumping during RNA synthesis. Many RNA viruses generate DI RNA and, thus, can potentially undergo genetic recombination. However, recombination has not been commonly observed. Therefore, there may be genetic restriction on RNA recombination or selective disadvantage for recombinant viruses.

RNA recombination in experimental systems with animal viruses

Evidence for RNA recombination was first obtained in classical genetic studies involving polioviruses (family Picornaviridae). When cells were co-infected with virus resistant to the antibodies in horse or cattle serum and virus resistant to guanidine, viruses resistant to both antibody and guanidine could be isolated at a higher frequency than by spontaneous mutation, suggesting that recombination may have occurred (Hirst, 1962; Ledinko, 1963). Similar observations were made on foot-andmouth disease virus (FMDV) (Pringle, 1965).

Introduction

A type of dynamic genetic organization of a population in which individual genomes have a fleeting existence, and usually differ in one or more positions from the consensus or average sequence of the population, is known as a quasi-species. Quasi-species have been found among viruses with RNA genomes and originate as a result of the high mutation rates during RNA replication and reverse transcription, as well as in the competition among continuously arising variant genomes. Because the concept of the quasi-species is of fundamental importance to understand RNA viruses, in this brief chapter we address the origins of the idea, summarize its biological relevance by referring the reader to other articles and reviews, and discuss the adequacy of the term quasi-species.

Historical background: the origins

The quasi-species concept had two origins: one theoretical and the other experimental.

The first theoretical treatment of a system involving regular generation of error copies by replication with limited fidelity was by one of the authors (Eigen, 1971) in considering earliest life forms on earth. This study was stimulated by Sol Spiegelman's experiments on in vitro replication of Qß RNA, and a ‘breakfast discussion’ with Francis Crick. ‘Self-instructive’ behaviour was distinguished from ‘general autocatalytic’ behaviour, the former being required for ‘template’ activity. A quality factor was defined which determines the fraction of copying processes that leads to an exact copy of the template. The ‘master copy’ will produce ‘error copies’ which will occur with a certain probability distribution.

Introduction

Influenza is the paradigm of a viral disease that relies on continued evolution of the virus to cause annual epidemics and occasional pandemics of disease in humans. The gene pool of influenza A viruses in aquatic birds provides all the genetic diversity required for the emergence of pandemic influenza virus for humans, lower animals and birds. In humans, pigs and horses, influenza A viruses show both antigenic drift and genetic shift (Webster et al., 1992). In contrast, there is emerging evidence that avian influenza viruses are in evolutionary stasis. This review reports that rapid evolution in influenza A viruses in humans and other mammals has continued since the beginning of recorded medical history and depends on periodic introductions of gene segments or entire influenza viruses from the avian influenza virus gene pool. In aquatic wild birds, influenza virus appears to be fully adapted to its host and causes no disease signs. Thus, understanding the ecology of influenza viruses in the reservoir of aquatic birds is essential if we wish to find ways to intervene and reduce or prevent the occasional catastrophic pandemics such as the one that decimated the human population of the world in 1918 after the appearance of ‘Spanish’ influenza.

Host-specific evolution of influenza viruses

It is likely that the NP protein gene determines host range (Scholtissek et al., 1985; Tian et al., 1985; Snyder et al., 1987); consequently NP gene evolution and host taxonomy has been studied (Gorman et al., 1990a, 1991; Gammelin et al., 1990).

Introduction and background

Point mutation has gained a lot of attention in molecular biology because of the implication of this genomic alteration in medicine and pathology, such as genetic disorders and cancers. The RNase A mismatch method was developed for the detection of point mutations related to the activation of the K-ras oncogene in colon tumours using RNA: RNA hybrids (Winter et al., 1985) and to the diagnosis of genetic disorders with RNA:DNA hybrids (Myers, Larin & Maniatis, 1985).

RNA viruses are characterized by great genetic variability. This implies the occurrence of frequent mutation in the genome of different isolates. Some of these mutations are involved in phenotypic properties, such as virulence, tropism, resistance to antiviral drugs and other characteristics (Domingo et al., 1985). They are also the basis for evolutionary studies and strain comparison. Mutations in RNA viruses have been detected by Tl oligonucleotide fingerprinting and lately by sequencing through cDNA.

Our group adapted the system of RNase A mismatch for studies on genetic variability of RNA viruses using influenza orthomyxovirus as a model (López-Galíndez et al., 1988) and Owen and Palukaitis (1988) to plant viruses. The system is based on the comparison by hybridization of a riboprobe from a reference strain with RNAs from different strains. Each one will give a complex pattern of bands resistant to the RNase A digestion which is specific for each one as a fingerprint. Comparing the different patterns we are able to draw a qualitative estimate of genetic relatedness and evolution of field strains (López-Galíndez et al., 1988, 1991).

Introduction

The Aphthovirus genus, family Picornaviridae, is composed of seven distinct serotypes which cause foot-and-mouth disease (FMD), a severe disease of cloven-hoofed animals. FMD has serious economic consequences (Pereira, 1981) and is enzootic in most South American and African countries, as well as in regions of Asia and the Middle East. Foot-and-mouth disease virus (FMDV) usually causes a systemic acute infection with high morbidity and low mortality (Shanan, 1962). In ruminants, the virus may also produce an asymptomatic, persistent infection that involves limited viral amplification (Van Bekkum et al., 1959). This type of infection has been proposed as an epidemiologically important reservoir of FMDV (Hedger & Condy, 1985).

The general structure and molecular features of FMDV are, in general, similar to those of other picornaviruses (Domingo et al., 1990; Stanway, 1990). The capsid is composed of 4 proteins (VP1-4) and includes an RNA molecule of about 8500 nucleotides in length which encodes the structural proteins and at least 11 different, mature, non-structural polypeptides. The antigenic structure of FMDV includes continuous and discontinuous neutralizing epitopes located in exposed regions of the viral capsid, in which one or more of the capsid proteins, particularly VP1, are involved (Domingo et al., 1990).

Aphthoviruses show considerable antigenic diversity; 7 serotypes, more than 65 subtypes and a multitude of variants have been identified mainly by in vitro cross-serum neutralization (Pereira, 1981) and, more recently, by the use of monoclonal antibodies (MAbs) (Domingo et al., 1990). Immunization with viruses of one type does not confer protection against viruses of other serotypes, whereas cross-protection within serotypes is not always complete (Kitching et al., 1989).

Introduction

Recombination of viral and host cellular nucleic acids probably plays a major role in the evolution of viruses. By this means, a virus is able to exploit a large gene pool and to acquire new properties very rapidly. The presence of cellular genes or gene fragments within the genomes of retroviruses (v-oncogenes) is well established. For other viruses, the cellular origin of certain genes has been suggested. Direct evidence is, however, difficult to obtain, probably because elaborate mutational changes of the cellular sequences occurred in the viral genomes. For some non-retro viruses, however, sequences homologous to cellular genes have been identified. We will discuss examples of these very recent recombination events with emphasis on bovine viral diarrhoea virus (BVDV).

Cellular sequences in the genome of BVDV

Identification of host cell-derived insertions

BVDV is a member of the genus pestivirus within the family Flaviviridae (Wengler, 1991). The pestiviral genome consists of single-stranded RNA with positive polarity of about 12 kilobases (kb) (Collett et aL, 1988; Meyers, Riimenapf & Thiel, 1989∧). The first third of this RNA codes for an autoprotease and four structural proteins, while the 3’ part of the genome encodes the other non-structural proteins (Collett et ai, 1991; Thiel et al., 1991). The genome of BVDV, like that of all flaviviruses, is translated into a single polyprotein which is co- and post-translationally cleaved to give rise to mature viral proteins.

Introduction

Brandes and Wetter (1959) first showed that potato virus Y together with 13 other viruses had filamentous flexuous virions about 750 nm in length, some of which were serologically related to one another. They proposed that these viruses formed a natural group, and this was subsequently named the potyvirus group by Harrison et al. (1971).

Large numbers of viruses have been added to the group, and it now contains at least 200 distinct species, or more than a fifth of all known plant viruses. It is the largest and most rapidly growing of the 50 or so families (or groups) of viruses that infect plants and is now named the Potyviridae comprising, at present, three genera, the potyviruses, rymoviruses and bymoviruses (Ward & Shukla, 1991; Barnett, 1992).

Poty viruses are found in all climatic zones, especially the tropics (Hollings & Brunt, 1981). They cause diseases in almost all crop plants and in many uncultivated species; in 1974 they were reported to infect 1112 plant species of 369 genera in 53 families (Edwardson, 1974), and that list too has grown. Their economic impact was also highlighted in a recent survey of important viruses with filamentous virions as 73% of those named were potyviruses (Milne, 1988). Thus it is important to try to understand how the potyviruses have evolved because it might give clues as to why they are such successful viruses.

Most potyviruses belong to the genus Potyvirus, and are transmitted in nature by aphids and through seeds of infected plants. These two properties, together with the diversity of crops they infect, assure the continuous presence of potyviruses in nature throughout the year (Hollings & Brunt, 1981).

Introduction

The evolutionary history of viruses, inferred from comparisons of genomic sequences, appears unlike that of most other life forms. Novel viruses seem often to arise through the conjunction of genes from very different viral lineages. Viruses with RNA genomes as well as viruses with DNA genomes have evolved in this way, and either homologous or non-homologous recombination may be involved. Botstein (1980) named this process modular evolution, and its importance is clear from evidence that, on several occasions, it has led to the evolution of new viral genera (Gibbs, 1987). However, little is known about the frequency of these events nor is it known if they occur in some pattern, and the interpretation of phylogenies disturbed by modular evolution is in its infancy.

In this chapter I present perhaps the first evidence of a pattern to viral modular evolution, and the first evidence of a connection between interviral gene transfer and an ecological interaction between viruses. This evidence comes from the luteovirus supergroup which consists of plant viruses with small messenger sense RNA genomes and icosahedral virions (Goldbach & Wellink, 1988; Martin et aL, 1990; Dolja & Koonin, 1991). Six viral groups and some ungrouped viruses belong to the supergroup (Brunt, Crabtree & Gibbs, 1990; Francki et ai, 1991). Many of the viruses are agriculturally important (Duffus, 1977; Matthews, 1991). The genomes of 17 of them have been completely sequenced. Analysis of this data shows the supergroup to consist of two main clusters that are in some ways only distantly related.

Introduction

Over the past half century, viruses have been classified and identified by such characters as the shape, size and serological specificity of their virions, and also by biological characters such as host and vector type. These features placed viruses into seemingly sensible genera and other taxa, but the evolutionary relationships of the genera (and higher taxa), and the processes that produced them, were unknown. However, knowledge of viral genomic sequences is starting to reveal the wealth of genetic processes involved in viral evolution, and the relative importance of each (Fig. 6.1). Sequence information has confirmed most of the earlier groupings, but also revealed a much greater shared genealogy than expected.

Taxonomic comparisons show that the genes of closely related viruses, such as isolates of a single species, usually only differ from one another by point mutations and, occasionally, by insertions or deletions of one or a few nucleotides. The different species of each viral genus differ by a greater number of changes of the same type. Thus mutation is a primary evolutionary driving force for viruses just as it is for all other organisms.

Comparisons of the genes of viruses placed in separate genera show that some are related, some are not. For example, all viral RNA replicase genes encode proteins that share sequence motifs, and hence may be related (Poch et al., 1989). Likewise, the virion proteins of most viruses with isometric virions of 25-30 nm diameter have a characteristic eight-stranded (3-barrel structure (Rossmann & Johnson, 1989) the only known exception being those of Q(3 levivirus, which has virion proteins with a (3-sheet/a-helical hook structure (Valegaord & Liljas, 1990).

This chapter briefly outlines the lines of evidence of virus evolution that are discussed in this book; hopefully it augments the information given in the Table of Contents by indicating where those lines anastomose. Evolution is the process whereby the population of an organism changes genetically over a period of time. It occurs when genetic variation within a population, combined with selection from among the mixture of genotypes, results in change. Viruses, especially those with RNA genomes, sometimes evolve very rapidly and so their evolution can be studied as it occurs. By contrast, all cellular organisms evolve more slowly and so their evolution is deduced by comparing extant forms and, for some, the fossils of their ancestors. No fossils of viruses are known.

Pre-molecular evidence of virus evolution

It has been realized for a long time that pathogens evolve. The fact that new epidemics of disease appear (Chapter 3), and that some viral diseases, such as measles and smallpox, induce life-long immunity in those individuals that survive, whereas others, such as the common cold or influenza, do not, all indicate that some viruses change. Of course, such differences influence choice of control strategies because stable viruses are amenable to control by vaccination, whereas rapidly changing viruses are not. Another type of evidence of virus evolution is the fact that avirulent strains of some viruses can be selected for use as vaccines, usually by growing them in unusual hosts, or under unusual conditions.

One of the first deliberate attempts to study natural virus evolution was the classic study of myxoma leporipoxvirus when it was liberated in Australia to attempt to control European rabbits (Chapter 2).

Introduction

Why biological control?

The past 15 years have witnessed a dramatic increase in research related to biological control. The current perception that biocontrol will have an important role in commercial agriculture in the future contrasts markedly with previously-held views that biocontrol agents perform too inconsistently, or are too narrow in their spectrum of activity, as compared with chemical pesticides, to be commercially feasible on a large scale. Renewed interest in biological control is in part a response to widespread concern about the potential negative impact of chemical pesticides on public health and the environment. Furthermore, the techniques of molecular biology have revolutionized the field by facilitating the identification of the molecular basis of pathogen suppression and by providing the means for construction of ‘superior’ biocontrol agents. New biocontrol agents resulting from recent intensive research are slowly becoming available to agriculture, and the trend should accelerate throughout this decade. One example is Gliocladium virens, which is being marketed in potting-mix to control Pythium and Rhizoctonia (see Lumsden and Walter, Chapter 25).

This chapter deals with the potential benefits and risks from the introduction of biocontrol agents for the control of root diseases of wheat, as well as the impediments to the application of this technology in commercial agriculture. The focus of the chapter is on biological control of take-all of wheat by fluorescent Pseudomonas spp. because it is a model system for the study of the molecular basis of pathogen suppression, root colonization by introduced bacteria and field application of biocontrol agents.