key: cord-022196-1tionxun authors: FENNER, FRANK; McAUSLAN, B.R.; MIMS, C.A.; SAMBROOK, J.; WHITE, DAVID O. title: The Nature and Classification of Animal Viruses date: 2013-11-17 journal: The Biology of Animal Viruses DOI: 10.1016/b978-0-12-253040-1.50006-3 sha: doc_id: 22196 cord_uid: 1tionxun nan Virology began as a branch of C H APT ER 1 pathology, the study of disease. At the end of the nineteenth century, when the microbial The Nature and Classification etiology of many infectious disof Animal Viruses eases had been established, pathologists recognized that there since been discovered, but two still apply: (a) unlike even the smallest microorganisms (chlamydiae), viruses contain no functional ribosomes or other cellular organelles, and (b) in RNA viruses the whole of the genetic information is encoded in RNA, a situation unique in biology. Other distinctions apply to some but not all viruses, e.g., the isolated nucleic acid of viruses of several genera is infectious (i.e., the virus can be generated intracellularly from a single molecule of nucleic acid), and viruses of most genera contain either no virus-coded enzymes, or one or more enzymes that belong to particular classes (neuraminidases and nucleic acid polymerases). It is impossible to define viruses satisfactorily in a sentence or even a paragraph, bearing in mind both their intracellular states and the extracellular particles or virions. Virions consist of a genome of either DNA or RNA enclosed within a protective coat of protein molecules, some of which may be associated with carbohydrates or lipids of cellular origin. In the vegetative state and as "provirus" (see Chapter 5), viruses may be reduced to their constituent genomes, and the simplest "viruses" may be transmitted from one host to another as naked molecules of nucleic acid, possibly associated with certain cellular components. At the other extreme, the largest animal viruses, e.g., the poxviruses and the leukoviruses, are relatively complex. Lwoff's concept that "viruses are viruses" has had important theoretical and practical consequences; on the one hand, it emphasized their similarities irrespective of the nature of the host (animal, plant or bacterium), and, on the other hand, it led to the possibility of freeing viruses from the rules of bacteriological nomenclature. However, the operational division of viruses made according to type of host continues to be used by the majority of virologists most of the time, and it is significant that the International Committee on Nomenclature of Viruses (ICNV), although dedicated to a universal classification, operates through Subcommittees on Bacterial, Invertebrate, Plant, and Vertebrate Viruses (Wildy, 1971 ). The simpler viruses consist of nucleic acid and a few polypeptides specified by it. More complex viruses usually also contain lipids and carbohydrates; in the great majority of viral genera these chemical components are not specified by the viral genome but are derived from the cells in which the viruses multiply. In exceptional situations, cellular nucleic acids or polypeptides may be built into viral particles. Viruses contain only a single species of nucleic acid, which may be DNA or RNA. Viral nucleic acid may be single-or double-stranded, the viral genome may consist of one or several molecules of nucleic acid, and if the genome consists of a single molecule this may be linear or have a circular configuration. 3 As yet, no animal viral nucleic acid has been found to be methylated, or to contain novel bases of the type encountered in bacterial viruses or mammalian transfer RNA's, but some virions contain oligonucleotides rich in adenylate, of unknown function. The base composition of DNA from animal viruses covers a far wider range than that of the vertebrates, for the guanine plus cytosine (G+C) content of different viruses varies from 35 to 74%, compared with 40 to 44% for all chordates. Indeed, the G+C content of the DNA of viruses of one genus (Herpesvirus) ranges from 46 to 74%. The molecular weights of the DNA's of different animal viruses varies from just over 1 to about 200 million daltons; the range of molecular weights of viral RNA's is much less, from just over 2 to about 15 million daltons. The nucleic acid can be extracted from viral particles with detergents or phenol. The released molecules are often fragile but the isolated nucleic acid of viruses belonging to certain genera is infectious. In other cases, the isolated nucleic acid is not infectious even though it contains all the necessary genetic information, for its transcription depends upon a virion-associated transcriptase without which multiplication cannot proceed. All DNA viruses have genomes that consist of a single molecule of nucleic acid, but the genomes of many RNA viruses consist of several different molecules, which are probably loosely linked together in the virion. In viruses whose genome consists of single-stranded nucleic acid, the viral nucleic acid is either the "positive" strand (in RNA viruses, equivalent to messenger RNA) or the "negative" (complementary) strand. Preparations of some viruses with genomes of single-stranded DNA consist of particles that contain either the positive or the complementary strand. Viral preparations often contain some particles with an atypical content of nucleic acid. Host-cell DNA is found in some papovaviruses, and what appear to be cellular ribosomes in some arenaviruses. Several copies of the complete viral genome may be enclosed within a single particle (as in paramyxoviruses) or viral particles may be formed that contain no nucleic acid ("empty" particles) or that have an incomplete genome, lacking part of the nucleic acid that is needed for infectivity. Terminal redundancy occurs in the DNA of some vertebrate viruses, but most sequences are unique. The largest viral genomes contain several hundred genes, while the smallest carry only sufficient information to code for about half a dozen proteins, most of which are structural proteins of the virion. The major constituent of the virion is protein, whose primary role is to provide the viral nucleic acid with a protective coat. As predicted by Crick and Watson (1956) , from a consideration of the limited amount of genetic information carried by viruses, the protein shells of the simpler viruses consist of repeating protein subunits. Sometimes the viral protein comprises only one sort of polypeptide chain, although, more commonly, there are two or three different polypeptides. The proteins on the surface of the virion have a special affinity for complementary receptors present on the surface of susceptible cells. They also contain the antigenic determinants that are responsible for the production of protective antibodies by the infected animal. Viral polypeptides are quite large, with molecular weights in the range 10,000-150,000 daltons. The smaller polypeptides are often but not always internal, the larger ones often but not always external. There are no distinctive features about the amino acid composition of the structural polypeptides of the virion, except that those intimately associated with viral nucleic acid in the "core" of some icosahedral viruses are often relatively rich in arginine. Viral envelopes usually originate from the cellular plasma membrane from which the original cellular proteins have been totally displaced by viral peplomers and a viral "membrane protein" (see Fig. 1-1) . The peplomers consist of repeating units of one or two glycoproteins, the polypeptide moiety of which is virus-specified while the carbohydrate is added by cellular transferases. In many enveloped viruses, the inside of the viral envelope is lined by a viral protein called the membrane or matrix protein. Not all structural viral proteins are primary gene products, since with many viruses the viral mRNA is translated into a large polypeptide that is enzymatically cleaved to yield two or more smaller virion proteins. Cleavage is often one of the terminal events in the assembly of the virion and it can occur in situ after most of the proteins are already in place. Although most virion polypeptides have a structural role some have enzymatic activity. Many viruses contain a few molecules of an internal protein that functions as a transcriptase, one of the two kinds of peplomers in the envelope of myxoviruses has neuraminidase activity, and a variety of other enzymes are found in the virions of the larger, more complex viruses. In addition to polypeptides that occur as part of the virion, a large part of the viral genome (most of it, with the large DNA viruses) codes for polypeptides that have a functional role during viral multiplication but are not incorporated into viral particles. Few of these "nonstructural viral proteins" have been characterized. Except for the large and complex poxviruses, which constitute a special case, lipid and carbohydrate are found only in viral envelopes and are always of cellular origin. The lipids of viral envelopes are characteristic of the cell of origin, though minor differences between the viral envelope and the normal plasma membrane may be demonstrable. About 50 to 60% of the lipid is phospholipid and most of the remainder (20-30%) is cholesterol. Some of the viral carbohydrate occurs in the envelope as glycolipid characteristic of the cell of origin, but most of it is part of the glycoprotein peplomers that project from the viral envelope. During the 4 years that followed the introduction of negative staining for the electron microscopic study of viruses (Brenner and Home, 1959), a general The Structure of Animal Viruses 5 picture was obtained of the structure of representatives of most of the groups of animal viruses that were known at the time (review: Home and Wildy, 1963). Three structural classes were distinguished: isometric particles, which were usually "naked" but in some groups were enclosed within a lipoprotein envelope; long tubular nucleoprotein structures, always (with viruses of vertebrates) surrounded by a lipoprotein envelope; and in a few groups, a more complex structure. Accepting a number of new terms defined by Lwoff et al. (1959a) , Caspar and his colleagues analyzed the principles underlying the structure of simple viruses (review: Caspar, 1965). Their basic concepts remain valid, but subsequent work has rendered some of the original definitions ambiguous; where necessary these have been modified. Virion (plural virions) is used as a synonym for "virus particle." The protein coat of an isometric particle or the elongated protein tube of viruses with helical symmetry is called the capsid. It may be "naked," or it may be enclosed within a lipoprotein envelope (peplos) which is derived from cellular membranes as the virus matures by budding. Where the capsids directly enclose the viral nucleic acid, as is usual with tubular capsids but less common with isometric capsids, the complex is called the nucleocapsid. With most isometric particles and in all complex virions, the capsid encloses another protein structure containing the viral genome, called the core. Capsids consist of repeating units of one or a small number of protein molecules. Three levels of complexity can be distinguished. Chemical units, the ultimate gene products, are single polypeptides that may themselves constitute the structural units, or several polypeptides may form homo-or heteropolymers which constitute the structural units. The structural units, or groups of them, may be visualized in the electron micrographs as morphological units. Morphological units that form part of a capsid are called capsomers; those projecting from the envelope are the peplomers (sometimes called "spikes," an unsatisfactory term since they are never pointed and may, indeed, have knob-shaped ends). The chemical units are sometimes held together by disulfide bonds to form the structural units, hence the practice of using reducing agents in polyacrylamide gel electrophoresis when analyzing viral proteins to determine their constituent polypeptides. The structural units are held together to form the capsid by noncovalent bonds, which may be polar (salt and hydrogen bonds) or nonpolar (van der Waals and hydrophobic bonds). The capsids of some viruses are readily disrupted in molar calcium or sodium chloride, suggesting electrovalent bonds between the structural units; others are unaffected by salt and can only be disrupted by detergents, suggesting that they are hydrophobically bonded. It has been found that the isometric virus particles that have been adequately studied by X-ray diffraction and electron microscopy have capsids in which the capsomers are arranged with icosahedral symmetry. According to Caspar and (B) . The capsids consist of morphological suhunits called capsomers, which are in turn composed of structural suhunits that consist of one or more chemical suhunits (polypeptide chains). Many icosahedral viruses have a "core" (not illustrated), which consists of protein(s) directly associated with the nucleic acid, inside the icosahedral capsid. In viruses of type B the envelope is a complex structure consisting of an inner virus-specified protein shell (membrane protein, made up of structural suhunits), a lipid layer derived from cellular lipids, and one or more types of morphological subunits (peplomers), each of which consists of one or more virus-specified glycoproteins (modified from Caspar et ah, 1962) . Klug (1962) , this occurs because the icosahedron is that polyhedron with cubic symmetry which, if constructed of identical subunits, would least distort the subunits or the bonds between them. A n icosahedron ( Fig. 1 -2) has 20 equilateral triangular faces, 12 vertices, where the corners of 5 triangles meet, and 30 edges, where the sides of adjacent pairs of triangles meet. It shows twofold symmetry about an axis through the center of each edge ( Fig. 1-2A) , threefold symmetry w h e n rotated around an axis through the center of each triangular face ( Fig. 1-2B) , and fivefold symmetry about an axis through each vertex ( Fig. 1-2C ). Each triangular face may be thought of as containing, and being defined by, three asymmetric units (i.e., units that have no regular symmetry axes themselves) so that a minimum of sixty asymmetric units are required to construct an icosahedron. The triangular faces of an icosahedron can be subdivided into smaller identical equilateral triangles, to form a solid called an icosadeltahedron. Only certain subdivisions are possible; the number of new triangles per facet is called the triangulation number (T), and T = h 2 + hk + k 2 , where h and k are any pair of integers. When h = k (T -3, 12, 27, etc.), or when either h or k = 0 (T = 1, 4, 9, 16, etc.), the triangles are arranged symmetrically on the underlying icosahedral face, but with other values for h and k (e.g., h = 2, k = 1 and T = 7) they are in a skew arrangement. A complete description then requires determination of the hand of the structure (right-dextro or left-Zevo). The hand of the icosahedral shells of some papilloma viruses has been investigated by Klug and Finch (1965) and Finch and Klug (1965) , who concluded that the human papilloma virus (human wart virus) had a T = 7d icosahedral surface lattice whereas rabbit papilloma virus had a T = 7l lattice. In an icosadeltahedron with a triangulation number of 3, each icosahedral face has 9 and the whole solid 2 0 X 9 = 180 asymmetric units. These structural units may differ in shape and clustering so that the morphological units (capsomers) visible by electron microscopy may differ greatly in viruses with the same triangulation number. There are three basic types of clustering pattern: 1. The three units defining each triangular face may cluster at the center of the triangle, forming trimer capsomers ( Fig. 1-2D ). 2. The structural units may cluster at the vertices of the triangles, so that where five triangles meet at the vertices of the icosahedron there are pentamer capsomers, and where six triangles meet on faces of the icosadeltahedron there are hexamer capsomers ( Fig. 1-2E) . 3. Pairs of structural units from adjacent triangles may cluster on the edges between the triangle to give dimer capsomers ( Fig. 1-2F ). The pattern seen on the surface of the virion need not reflect the way in which the structural units are bonded together, and gives no clue as to whether the structural units are constituted by single chemical units or are homo-or heteropolymers of the chemical units. However, the number of structural units in each capsomer can be guessed at from the arrangement and size of the capsomers ( Fig. 1-2) . All known animal viruses whose genome is DNA have isometric (or complex) capsids, as do all those whose genome is double-stranded RNA and the viruses of two major families (Picornaviridae and Togaviridae) whose genome consists of a single molecule of single-stranded RNA. Tobacco mosaic virus occupies a unique position in virology. Not only was it the agent whose "viral" nature was first appreciated (Beijerinck, 1899), and the first virus to be crystallized (Stanley, 1935), but more is known of its physical and chemical structure than any other virus (reviews: Caspar, 1965; Kaper, 1968). The virus particles are nonenveloped straight rods, which consist of 2100 repeating polypeptide (chemical) units, which are the structural units and also, without clustering, constitute the capsomers. These protein molecules are arranged in a helical manner so that except at the ends of the particles every capsomer is in a structurally equivalent position in relation to the long axis of the rods. Many plant viruses and a few bacteriophages have similar nonenveloped tubular virions, their capsomers being arranged in helices whose pitch is characteristic for the virus group. Such viruses are structurally defined by their length and width, the pitch of the helix, and the number of capsomers in each turn of the helix. Tubular nucleocapsids are found in many groups of viruses of vertebrates, but only among those whose genome consists of single-stranded RNA. None of these occurs as "naked" virions; the flexuous helical tubes are always inside lipoprotein envelopes. The diameters of the nucleocapsids of several viruses have been measured, but in only a few cases is the length or the pitch of the helix known. The best studied example, the nucleocapsid of Sendai virus, a paramyxovirus, is a helix about 1 /xm long and 20 nm wide, with a pitch of 5.0 nm (Finch and Gibbs, 1970) . There are about 2400 hourglass-shaped structural units in the nucleocapsid, with either eleven or thirteen units per turn of the helix. The structural units are single polypeptides with a molecular weight of about 60,000 daltons, arranged with their long axes at an angle of about 60° to the long axis of the nucleocapsid, which therefore has a herringbone appearance in electron The Structure of Animal Viruses 9 micrographs (see Plate 3-16). The contact surface between adjacent turns of the basic helix is conical, so that contact is maintained even when the nucleocapsid is sharply flexed, and the viral RNA is thus protected. Unlike cells, which contain several different species of nucleic acid that subserve different functions, the only nucleic acid in viruses, apart from small amounts of host fRNA in leukoviruses, is their genome. It may consist of either DNA or RNA, it may be single-or double-stranded, it may be linear or cyclic, and the genome may consist of one or several molecules of nucleic acid. Although the only detailed studies have been made on a few plant and bacterial viruses (review: Tikchonenko, 1969), it is clear that the interaction between the viral nucleic acid and the capsomers is different in nucleocapsids with helical and icosahedral symmetry. In tobacco mosaic virus, there is a maximum regular interaction between the single strand of viral RNA and the protein subunits which form a protective coat around it. A similar relationship probably exists in most animal viruses with tubular nucleocapsids, but in some viruses (e.g., influenza virus) the integrity of the tubular structure is destroyed by treatment with RNase but not by proteases, suggesting a different relationship of RNA and protein. In icosahedral viruses, on the other hand, there can be no such regular relationship of the nucleic acid and each polypeptide subunit. In the simplest isometric viruses, the folding of the flexible single-stranded RNA may have some regularity in relation to the capsomers and their constituent chemical subunits. X-ray diffraction studies of turnip yellow mosaic virus (Klug et al., 1966), for example, show that a significant portion of the single RNA chain is deeply embedded within the protein shell, large segments being intimately associated with the 180 structural units, which as hexamers and pentamers make up the 32 capsomers. The presence of the RNA in and about these positions enhances the definition of 32 capsomers seen in electron micrographs (Finch and Klug, 1966) . Except for some togaviruses, even the simple isometric viruses of vertebrates have a more complex structure than this, since they contain several virus-coded poly pep tides. One or more of these poly pep tides are known to be "internal" to the capsid and it is thought that these rather than the capsomers interact with the viral RNA. Reovirus particles have two concentric protein shells, each consisting of well-defined morphological units. The proteins of the larger DNA viruses are arranged in several layers, not all of which display symmetry. The internal proteins of many DNA viruses are highly basic and are thought to be bonded to the viral nucleic acid, constituting a core within the isometric capsid. Although occasionally used in a more general way to refer to the outer viral coats of some complex viruses like the poxviruses (Mitchiner, 1969), we think that it is desirable to restrict the use of the term "envelope" to the outer lipoprotein coat of viruses that mature by budding through cellular membranes. Enveloped viruses contain 20-30% of lipid, all of which is found in the envelope. Chemical analyses show that the lipid is derived from the cellular membranes through which the virus matures by budding, but all the polypeptides of viral envelopes are virus-specified. Herpesvirus is the only virus of vertebrates that matures by budding through the nuclear membrane, and its envelope contains several virus-specified glycoproteins. All other enveloped viruses bud through cytoplasmic membranes, and contain one or more different polypeptides. The Togaviridae have an isometric core to which a lipid layer is directly applied, and virus-specified glycoprotein peplomers project from this. All animal viruses with tubular nucleocapsids are enveloped, and in these the lipid layer from which glycoprotein peplomers project is probably applied to a protein shell (the membrane protein; see Fig. 1 -1), which may be relatively rigid, as in Rhabdovirus, or readily distorted (as in the myxoviruses) so that in negatively stained electron micrographs the virions appear to be pleomorphic. Viruses that have large genomes have a correspondingly complex structure. Apart from the undetermined nature of the "cores" of many of the isometric viruses (e.g., Herpesvirus and Adenovirus), the virions of the two largest animal viruses (Poxvirus and Iridovirus) have highly complex structures, which are described in the appropriate sections of Chapter 3. The RNA viruses that have the largest (single-stranded) genomes, those of the Leukovirus genus, also have a highly complex structure with an envelope enclosing an icosahedral capsid that, in turn, surrounds a tubular nucleocapsid. The aim of classification in biology is to make an ordered arrangement of a particular class of biological objects that will indicate their similarities and differences. Adoption of a system of classification also involves consideration of the nomenclature of the objects to be classified. Linnaeus introduced a latinized binomial nomenclature into biology 200 years ago, and phylogenetic classifications of animals and plants based on the theory of evolution have since been introduced. International Codes of Nomenclature with rigid sets of rules, and Judicial Commissions to pass judgement on proposed names, have been set up for the naming of plants and of animals. An International Code of Nomenclature of Bacteria and Viruses was approved in 1947 and has since been revised (Buchanan et ah, 1958). Although they are primarily concerned with nomenclature, all these Codes involve agreement upon a system of classification. Codes are based on "acceptances," i.e., beliefs we would like to justify but are unable to prove, the principal one being that we are able to arrange living things in an orderly system that is indicative of both rank in a hierarchy and phylogenetic relationships (Cowan, 1966) . Classifications of animals and plants attempt to be scientific by deriving their taxa from a consideration of phylogenetic relatedness. More recently this approach has been reinforced by tests for genetic relatedness, i.e., the information content of the genetic material of the agents concerned. This has been tested by homology experiments with DNA's ex- tracted from the cells of a variety of animals (McCarthy, 1969) , and it is to be expected that the phylogenetic and the molecular biological approaches will eventually be combined. The classification of bacteria into the same hierarchical pattern as that of plants and animals (phyla, subphyla, classes, orders, suborders, families, genera, and species) has led to a chaotic situation (Cowan, 1970) . Some bacterial taxonomists are looking to numerical methods, readily exploited with the aid of electronic computers, for the solution of their problems (Sneath, 1964) . Disadvantages of this approach are that the weighting of characters tends to be involuntary, and that pleiotropism may lead to some characters being scored more than once. Most virologists believe that certain characters of viruses, such as the type, amount, and conformation of the viral nucleic acid, are taxonomically more important than characters like host range or pathogenic potential. Molecular biology provides an alternative to phylogenetic relationships for making a scientific classification of microorganisms, viz., by the determination of genetic relatedness, using both the genetic material and the polypeptides that it specifies (Mandel, 1969) . There are two groups of agents, the mycoplasmas and the viruses, for which detailed "official" classifications are still in the process of formation. Because of their small genomes, they are particularly suitable for molecular taxonomy, i.e., classification based on the molecular weights and base ratios of their genomes, and on the results of nucleic acid hybridization experiments. Applied to mycoplasmas, this approach has disproved claims that these microorganisms were derived from certain bacterial species (Razin, 1969) . Nucleic acid hybridization experiments have now been performed with many different viruses; detailed references to the results obtained will be given in Chapter 3. In general, they have provided some useful data on relationships within genera and species, but not at higher taxonomic levels. With the methods used thus far many viruses now allocated to the same genus have shown little or no homology of their nucleic acids. Indeed, nucleic acid hybridization may be too critical a method to be useful except for the comparison of closely related viruses, and less exacting tests for the similarity of viral genomes may be more pertinent when considering different viral species. Bellett (1967a,b) analyzed the data available in 1966 on the molecular weights and base ratios of the nucleic acids of different viruses. His results on the "clustering" of the viruses of vertebrates are consistent with the genera proposed by the ICNV (Wildy, 1971). However, newer knowledge about the fragmented nature of the genomes of some RNA viruses and of the varied modes of their transcription and translation (Baltimore, 1971b) suggests that these data, where available, should be added to the parameters used by Bellett. The differences between the molecular weights of the DNA's of the DNA viruses of vertebrates are such that sophisticated analysis is not needed to define the currently accepted families and genera. Until about 1950, little was known about viruses other than their pathogenic behavior. Most early proposals for viral classification were confined to either plant or animal viruses and were based mainly upon the symptomatology of diseases caused by them, which tended to classify the host responses rather than the viruses. Bawden (1941) made the pioneering suggestion that viral nomenclature and classification should be based upon properties of the virus particle. In the early 1950's Bawden's approach was exploited by animal virologists (Andrewes, 1952), and viruses were allocated to groups which were usually given latinized names constructed from a chosen prefix plus the word "virus." Thus, myxovirus (Andrewes et ah, 1955) Group, 1963) , and adenovirus (Pereira et ah, 1963) groups were described. In the meantime, a classification using quite different criteria had been established by epidemiologists. Since they were so concerned with the transmission of infection, epidemiologists have used a classification based on the mode of transmission of disease; they have grouped viruses together as "respiratory viruses," "enteric viruses," or "arthropod-borne (arbo-) viruses." The last term, in particular, has been widely used, but it is generally agreed that this epidemiological classification, although useful is in no sense taxonomic. Concurrently with these suggestions relating to the viruses of vertebrates, Lwoff (1957) insisted upon the similarities between viruses, whatever their natural host, and the differences between viruses and all other biological entities. He was instrumental in arranging for the establishment of an international committee (Anon., 1965; Lwoff and Tournier, 1966) to discuss nomenclature. Its major proposal was to select "type species" upon which names for groups would be based. It also proposed a classification based on (a) the chemical nature of the nucleic acid, (b) the symmetry of the nucleocapsid (helical, cubical, or binal), (c) the presence or absence of an envelope, and (d) certain measurements: for helical viruses, the diameter of the nucleocapsid, for cubical viruses, the triangulation number and the number of capsomers. The official International Committee on Nomenclature of Viruses (ICNV), which was set up at the Ninth International Congress for Microbiology in 1966, adopted the physicochemical criteria of Lwoff and Tournier, but rejected the detailed hierarchical classification. The more important nomenclatural proposals accepted by ICNV were: (a) an "effort should be made" toward a latinized binomial system of nomenclature, (b) the "law of priority" is unacceptable, (c) no taxon should be named from a person, and (d) anagrams, siglas, hybrids of names, and nonsense names should be prohibited. Before describing the classification of animal viruses that we shall use throughout this book, it is appropriate to consider some of the problems of classification and nomenclature that have not yet been tackled by ICNV. One of the most important is the level of taxa that should be used. So far, only three families of animal viruses have been accepted (see below), but it is clear that large and heterogeneous groups currently classed as genera (e.g., Poxvirus, Herpesvirus, Paramyxovirus, Leukovirus; Wildy, 1971) should be regarded as families. Indeed, it would not be unreasonable to regard all the currently accepted isolated "genera" as families, some of which (e.g., Adenovirus) might at this A Classification of Animal Viruses 13 stage contain only a single genus. The conventional physicochemical criteria [(a) nucleic acid: type, strandedness, fragmentation, and molecular weight; (b) virion: shape, size, and symmetry] are suitable for classification at this level of family/genus, perhaps assisted by the serological cross-reactivity of "group" antigens where these have been recognized. At the other end of the nomenclatural spectrum, there is hopeless confusion in the ways in which the terms "species," "type," "subtype," and "strain" are used. For example, "types" of influenza virus exhibit no serological crossreactivity and their nucleic acids do not hybridize; they should be regarded at least as distinct species. On the other hand, many alphaviruses and flaviviruses with distinct names, which exhibit extensive serological cross-reactivity, should perhaps be regarded as types within the same species. Serological cross-reactivity and nucleic acid hybridization tests are probably most useful for making comparisons at this "species" level. The ICNV, working under the chairmanship of Professor P. Wildy, presented its first report at the Tenth International Congress for Microbiology in Mexico City in 1970, and has published valuable basic data on forty-three viral groups encompassing viruses of bacteria, invertebrates, plants, and vertebrates (Wildy, 1971) . In spite of the problems referred to above, we shall follow the classification set out in the Report, amplifying it with proposals that have come forward since then, but maintaining accepted usages of the terms "type," "strain," etc. Only two families were established by ICNV in 1970 (Wildy, 1971): Papovaviridae and Picornaviridae, and subsequently the family Togaviridae was defined. Most accepted "groups" of vertebrate viruses were given generic names. No species names were adopted by ICNV, although "t> ~>e species" were designated for several of the genera. The family Papovaviridae (sigla: Pa = papilloma; po = poly oma; va = vacuolating agent, SV40) encompasses two genera, Polyomavirus (poly = many; oma = tumor) and Papillomavirus (papilla ^nipple; oma = tumor), which differ substantially in size and nucleic acid content of the virion (Table 1 -2) but share many other properties. An important property of many papovaviruses is their capacity of produce tumors. In nature, some produce single benign tumors (which may undergo malignant change) and are highly host specific ; others may cause primary malignant tumors within a short period of their inoculation into newborn rodents. The adenoviruses (adeno = gland) are nonenveloped icosahedral DNA viruses which multiply in the nuclei of infected cells, where they may produce a crystalline array of particles. Many serological types have been isolated from human sources. These have an antigen that is shared by all mammalian strains, but differs from the corresponding antigen of avian strains. Allocation to the genus is made primarily on the basis of the characteristic size and symmetry of the virion as seen in electron micrographs (icosahedron with 252 capsomers). Most adenoviruses are associated with respiratory infection and many such infections are characterized by prolonged latency. Some multiply in the intestinal tract and are recovered in feces. Many adenoviruses, from both mammalian and avian sources, produce malignant tumors when inoculated into newborn hamsters. In the laboratory, stable hybrids have been produced between certain adenoviruses and the Polyomavirus, SV40 (see Chapter 7). The herpesviruses (herpes = creeping) are readily recognized by their morphology. Their icosahedral capsid is assembled in the nucleus and acquires an envelope as the virus matures by budding through the nuclear membrane. Electron microscopic examination by negative staining of many previously unclassified viruses showed that several of them had large icosahedral capsids with 162 capsomers enclosed within lipoprotein envelopes, similar to the type species, herpes simplex virus. When examined further, such viruses were found to be DNA viruses that multiplied in the nucleus, and have now been included in the genus Herpesvirus. Table 1 -4 shows some of the viruses now regarded as members of this genus; a more complete list is given by Andrewes and Pereira (1972). There is a group-specific antigen(s) associated with the nucleocapsids and demonstrable by immunodiffusion, and several type-specific antigens associated with the nucleocapsid and envelope. Some type-specific antigens crossreact (e.g., herpes simplex viruses type 1 and type 2 and B virus). Different herpesviruses cause a wide variety of types of infectious diseases, some localized and some generalized, often with a vesicular rash. A feature of many herpesvirus infections is prolonged latency associated with one or more episodes of recurrent clinical disease. This genus (irido = iridescent) was defined on the basis of several viruses of insects whose structure and nucleic acid content have been carefully studied 18 1. Nature and Classification of Animal Viruses (Bellett, 1968; Wrigley, 1969) . Several DNA viruses of vertebrates that are similar in morphology and certain other characteristics have been tentatively grouped with the genus Iridovirus (Table 1-5) . Like poxviruses, but unlike other DNA viruses, iridoviruses multiply in the cytoplasm. Their DNA consists of a single linear molecule, with a molecular weight of about 130-140 million daltons, and the virion is a large and complex nonenveloped icosahedron, with an outer shell composed of about 1500 capsomers. The vertebrate ''iridoviruses" may be enveloped. Several enzymes are found within mature virions. The best studied of the vertebrate "iridoviruses" are some viruses of frogs, notably FV3 (review, Granoff, 1969); the most important economically is African swine fever virus (review, Hess, 1971). The poxviruses (pock = pustule) are the largest animal viruses, and contain a larger amount of DNA (160-200 million daltons of double-stranded DNA) than any other virus. The structure of the brick-shaped virion is complex, consisting of a biconcave DNA-containing core surrounded by several membranes of viral origin. There is a poxvirus group antigen which is probably an internal component of the virion, and can be demonstrated by complement fixation or gel diffusion tests. Several enzymes, including a transcriptase, are found within mature virions. Multiplication occurs in the cytoplasm and the virions mature in cytoplasmic foci. Occasionally, the virion may be released within a loose membrane derived from the cytoplasmic membrane. This is not essential for infectivity, and must be distinguished from the envelope of viruses that mature by budding through cellular membranes. The genus is divided into several subgenera (Table 1-6) , and there are several poxviruses that have still to be classified. The properties outlined for the genus apply to all the subgenera, except that the virions of members of the subgenera B and C (see Table 1 -6), and swinepox virus, are narrower than those of other poxviruses, and virions of subgenus B (orf) have a distinctive surface structure. Species within each subgenus show a high degree of serological cross-reactivity by neutralization as well as complement fixation tests. Genetic recombination occurs within, but not between, subgenera; nongenetic reactivation (complementation) occurs between most poxviruses of vertebrates (Chapter 7). Certain viruses that multiply in insects have many of the attributes of poxviruses and have been tentatively called entomopoxviruses (entomo = insect) (review: Bergoin and Dales, 1971). Poxviruses cause diseases in man, domestic and wild mammals, and birds. These are sometimes associated with single or multiple benign tumors of the skin, but are more usually generalized infections, often with a widespread vesiculo-pustular rash. Several poxviruses are transmitted in nature by arthropods acting as mechanical vectors. The viruses of rodents cause acute fulminating disease when inoculated into newborn hamsters (Kilham, 1961) . The adenovirus-associated viruses are not known to cause any symptoms. The picornavirus group (sigla: pico = small; rna = ribonucleic acid), which includes a very large number of viruses, was accepted by ICNV as a family, Picornaviridae, with three genera: Enterovirus (entero = intestine), Rhinovirus (rhino = nose), and Calicivirus (calici = cup). Newman et al. (1973) believe that this subdivision of the family is unduly restrictive; on the basis of particle density, base composition of the viral RNA's, and stability at various pHs they differentiate cardioviruses (cardio = heart) from Enterovirus, and foot-and-mouth disease virus and "equine rhinovirus" from the genus Rhinovirus (Table 1- primarily inhabitants of the intestines, and a large number of serotypes have been found in the feces of man and of various animals. The enteroviruses of man have been subdivided into three major subgroups : poliovirus, three serotypes; echovirus (acronym: echo = enteric cytopathogenic /luman orphan), thirty-four serotypes; and coxsackievirus (Coxsackie = town in New York State), twenty-four serotypes of type A and six of type B. The polioviruses, which show some serological cross-reactivity, are distinguished by their capacity to paralyze humans. Coxsackieviruses were originally defined in terms of their capacity to multiply in infant mice, but subsequently some echoviruses were found to do the same. It has been recommended that all future en-Short Descriptions of the Major Groups of RNA Viruses 23 teroviruses that are discovered should be numbered sequentially from 68, irrespective of subgroups (Rosen et al., 1970) . Most infections with enteroviruses are inapparent, a few are associated with gastrointestinal disorders, and some may cause generalized infections with rash, central nervous system involvement, including poliomyelitis and aseptic meningitis, or specific damage to the heart. Genus: Rhinovirus [R/1: 2.6-2.8/30: S/S: V/0]. The rhinoviruses resemble the enteroviruses in several characteristics but they are acid labile (pH 3) and have a buoyant density (in CsCl) of 1.38-1.43 g/cm 3 . Most have a low ceiling temperature of growth and are characteristically found in the upper respiratory tract of man and various animals. There are a large number of different serotypes of human rhinoviruses, and there are several serotypes of foot-and-mouth disease virus, which resemble rhinoviruses in some respects, but not in others (Table 1-9) . Most rhinoviruses cause mild localized infections of the upper respiratory tract, but foot-and-mouth disease virus causes a severe generalized disease with rash in cattle. During the last quarter century intensive world-wide efforts have been made to recover viruses which would multiply in both arthropods and vertebrates, and some 200 different agents with these biological properties are now known. They have been called "arthropod-borne viruses," a name which was shortened to "arborviruses" and then (in order to avoid the connotation of "tree") to "arboviruses." The arboviruses have been defined, on epidemiological grounds (mode of transmission), as a group comparable to the "respiratory viruses." Arboviruses are viruses which, in nature, can infect arthropods that ingest infected vertebrate blood, can multiply in the arthropod tissues, and can then be transmitted by bite to susceptible vertebrates (World Health Organ., 1961). For many years arboviruses have been recovered from vertebrate tissues and suspensions of arthropods by the intracerebral inoculation of mice, and advantage has been taken of certain chemical and physical properties found to be commonly associated with them to avoid confusion with murine picornaviruses. The property generally tested was sensitivity to lipid solvents. Many arboviruses have lipoprotein envelopes and their infectivity is destroyed by these reagents (Theiler, 1957; Casals, 1961) . There was thus a tendency to equate sensitivity to lipid solvents with "arbovirus." During the last decade it has been recognized that the arbovirus group is quite heterogeneous in its physicochemical properties (see Table 16 -3). Some members are not enveloped (Orbivirus, Nodamura virus), and those sensitive to lipid solvents belong to at least three major groups (Togaviridae, Rhabdovirus, and Bunyamwera supergroup). This preamble has been necessary because in the past the term "arboviruses" has been regarded as applying particularly to viruses with the physicochemical properties of the group A and group B arboviruses. These viruses now form two genera (Alphavirus and Flavivirus) of the family Togaviridae (toga = cloak). (Table 1 -10) and show serological cross-reactivity by the hemagglutinin-inhibition test. The arthropod vectors are mosquitoes, but some alphaviruses may be transmitted congenitally by vertebrates. In nature, they usually cause inapparent infections of birds, reptiles, or mammals, but some can cause generalized infections associated with encephalitis in man and in other mammals. Genus: Flavivirus [R/1: 4/7-8: S/S: V,l/0, Di, Ac]. This genus (flavi = yellow) comprises the group B arboviruses. All members show serological crossreactivity. The arthropod vectors may be ticks or mosquitoes, and some of them may be transmitted by the ingestion of contaminated milk. They differ from the alphaviruses in that budding usually occurs into cytoplasmic vacuoles rather than from the plasma membrane. Most cause inapparent infections in mammals and less commonly in birds, but generalized infections of man may occur with visceral symptomatology (e.g., yellow fever), rashes (e.g., dengue), or encephalitis (e.g., Japanese encephalitis). Other Possible Members of the Family Togaviridae. On the basis of the physicochemical definition proposed, several other viruses that are not transmitted by arthropods should probably be included in this family. Generic names have not yet been proposed for these viruses, which include rubella and equine arteritis viruses, and the two serologically related viruses of hog cholera and bovine mucosal disease. In early classifications, some members of two very different genera, now distinguished from each other as Orthomyxovirus (ortho = correct; myxo = mucus) and Paramyxovirus, were grouped together as Myxovirus (Andrewes et ah, 1955). The common properties were an RNA genome, a tubular nucleocapsid, and a pleomorphic lipoprotein envelope that carried the properties of hemagglutination and enzymatic elution. The term "myxovirus" is now only used as a vernacular expression to encompass the viruses that have these properties {viz., influenza, mumps, Newcastle disease, and parainfluenza viruses); it has no taxonomic status. Type A influenza viruses have been recovered from a number of different species of animal (birds, horses, and swine) as well as man; types B and C are specifically human pathogens. They are an important cause of respiratory disease in man and other animals, and some of the avian influenza viruses may cause severe generalized infections. In contrast to the orthomyxoviruses, the paramyxoviruses (para = alongside; myxo = mucus) are enveloped viruses whose RNA occurs as a single linear molecule with a molecular weight of about 7 million daltons (Table 1-12) . The tubular nucleocapsid has a diameter of 18 nm and is about 1.0 μπι long. It is enclosed within a pleomorphic lipoprotein envelope 150 nm or more in diameter; long filamentous forms with the same diameter also occur. Three serologically related viruses, those of measles, distemper, and rinderpest, have been tentatively allocated to the Paramyxovirus genus on the basis of the morphology of the virion and nucleocapsid; they do not have a neuraminidase; respiratory syncytial virus is different again. Some paramyxoviruses cause localized infections of the respiratory tract and several produce severe generalized diseases; among the latter some are characteristically associated with skin rashes. The genus Arenavirus (arena = sand) was defined in terms of the electron microscopic appearance of the virions in thin sections, and serological crossreactivity (Rowe et al., 1970a) . The pleomorphic enveloped virions are 85-120 nm in diameter (sometimes larger), and have closely spaced peplomers. The structure of the nucleocapsid is unknown, but in thin sections the interior of the particle is seen to contain a variable number of electron-dense granules 20-30 nm in diameter, hence the name. All members of the genus are associated with chronic inapparent infections of rodents; some cause acute generalized diseases in other hosts (e.g., Lassa fever virus in man). b Characteristics : single-stranded RNA probably in several pieces, total molecular weight 3.5 million daltons; lipoprotein envelope 85-300 nm in diameter; multiply in cytoplasm; mature by budding from plasma membrane. All members share a group-specific antigen. Envelope encloses "granules" 20-30 nm in diameter; some of these are cellular ribosomes. The Bunyamwera "supergroup" of arboviruses (Bunyamwera, a locality in Africa) was established by Casals (World Health Organ., 1967) to bring together a number of minor arbovirus groups linked by distant serological reactions between occasional "bridging" viruses. The subgroups of viruses included are shown in Table 1 -15. All these viruses, numbering well over 100, are known or suspected to be arthropod-borne. Morphologically, those that have been studied have enveloped roughly spherical virions 90-100 nm in diameter with a tubular nucleocapsid. Several other arboviruses serologically unrelated to those of the "supergroup" have a similar morphology (Table 1 -15). Their genome consists of single-stranded RNA probably occurring in several pieces; its molecular weight has not been determined. The outstanding characteristic of the genus Leukovirus (leuko = white) is that all members contain an RNA-dependent DNA polymerase ("reverse transcriptase"). The viruses contain three or four pieces of single-stranded RNA, with a total molecular weight of 10 to 12 million dal tons, associated with a helical nucleocapsid, which is enclosed within a capsid with cubic symmetry. This is, in turn, enclosed within a lipoprotein envelope about 100 nm in diameter, containing peplomers which confer the type specificity. Leukoviruses mature by budding from the plasma membrane. Genome is a linear molecule of single-stranded RNA, 10-12 million daltons molecular weight, consisting of three to four linked pieces and probably associated with tubular nucleocapsid. Structure of virion is complex, the nucleocapsid being enclosed within a capsid of cubic symmetry, which is enclosed in an envelope that carries type-specific antigens. Virion also contains species-specific (e.g., feline or murine) and interspecies-specific (e.g., avian or rodent) antigens. As Table 1 -16 illustrates, the genus Leukovirus accepted by ICNV is clearly an inadequate taxon for the variety of viruses that now fulfill the physicochemical criteria set out above. The term "oncornaviruses" (Nowinski et al., 1970) is also not suitable for the taxon as a whole or any subgroup of it, for not all the viruses conforming to the physicochemical specifications of the genus Leukovirus are tumor viruses, many "RNA tumor viruses" are not transforming (Temin, 1972) , and in any case the leukemia-sarcoma viruses and the mammary tumor virus (both of which produce tumors) belong to different subgenera. In order not to prejudge a classification of the group, we shall adhere to the accepted generic name, but indicate four subgenera. Subgenus A includes the "C-type particle" viruses, some of which cause leukemia-sarcoma and are currently being subjected to intensive study. The mammary tumor virus, which is the best known representative of the subgenus B, differs from viruses of subgenus A in morphology and maturation ("B-type particles") and shows no serological cross-reactivity with the murine viruses of subgenus A. Subgenus C includes a group of serologically related viruses that cause slowly progressive diseases in sheep. They have all the physicochemical properties of leukoviruses. Although they do not cause neoplastic disease, they will transform cells that are nonpermissive for viral growth (Takemoto and Stone, 1971). The viruses of subgenus D (foamy agents) include a number of viruses of monkeys, cats, and cattle, that have no known pathogenic potential but have been frequently isolated from tumors (as "passenger viruses") or healthy animals. They have a different morphology from other leukoviruses (Clarke et al., 1969) and produce an intranuclear antigen as well as cytoplasmic antigens in infected cells (Parks and Todaro, 1972), but they contain a reverse transcriptase and are much more resistant to UV irradiation than other RNA viruses. The rhabdoviruses (rhabdo = rod) are enveloped RNA viruses with singlestranded RNA of molecular weight 4 million daltons. The RNA is associated with a very regular double-helical nucleocapsid 5 nm in diameter, enclosed within a bullet-shaped shell that measures about 175 X 75 nm (Table 1-17) . Several arboviruses belong to this genus, which also includes rabies virus and the virus of hemorrhagic septicemia of trout. It has been claimed that rabies virus can be adapted to multiply in Drosophila melanogaster (Plus and Atanasiu, 1966). Several viruses with a somewhat similar morphology cause diseases of insects and plants (Table 1-17, and see Table II of Howatson, 1970), but it may well turn out that these resemblances are superficial. Examination of the nature of their genomes and polypeptides is necessary before it can be confidently stated whether these viruses rightly belong to the genus Rhabdovirus or even to an enlarged family that might be called Rhabdoviridae. Nature and Classification of Animal Viruses The mammalian serotypes share a common antigen, which differs from the group antigen of the avian serotypes.Clover wound tumor virus, which multiplies in plants and leafhoppers, resembles the reoviruses of vertebrates morphologically and chemically but does not cross-react with them serologically. Bluetongue virus, an arbovirus, was found to resemble the reoviruses in some properties but not in others (review: Howell and Verwoerd, 1971 ). Subsequently, a large number of similar viruses have been recognized (Table 1 -19) and the name "Orbivirus" (orbis = ring) was suggested for them (Borden et a\., 1971 ). Reoviruses and orbiviruses may eventually be grouped together in the same family for which the name "diplornavirus" has been suggested (Verwoerd, 1970) . Apart from its "illegality" (according to ICNV Rules), the occurrence of other quite different viruses with genomes of double-stranded RNA (like some of the viruses of fungi and insects) cautions against ready acceptance of this term.All members of the genus multiply in arthropods as well as vertebrates. Some of them (bluetongue and Colorado tick fever viruses) cause severe generalized diseases with viremia in some vertebrates. It is pleasing to note that several of the viruses listed as "unclassified" in the first edition of this book have now been allocated to genera (lymphocytic choriomeningitis of mice, Arenavirus; mouse hepatitis virus, Coronavirus; rubella virus, Togaviridae; visna virus, Leukovirus; and African swine fever virus, Iridovirus). A few unclassified viruses remain that warrant special mention here, such as the human hepatitis viruses, the agents of the subacute spongiform encephalopathies (scrapie, etc.), lactic dehydrogenase elevating virus (LDV), and the Marburg agent. Experiments with human volunteers many years ago (Neefe et al, 1945) and again more recently (Krugman et al., 1967) have shown that the diseases commonly known as infective hepatitis and serum hepatitis are caused by two viruses that differ serologically, in their clinical expression, and in their usual routes of transmission. Because both can be transmitted orally it is better to use noncommital names for them, and "serum hepatitis" is now termed hepatitis B; infective hepatitis, hepatitis A. Study of these viruses has been greatly inhibited by the lack of susceptible laboratory animals (chimpanzees may get clinical hepatitis; marmosets and rhesus monkeys subclinical infection, while other laboratory animals are insusceptible), and the difficulty of obtaining reproducible cytopathic changes in cultured cells. The recognition in the sera of cases of serum hepatitis of lipoprotein particles of characteristic serological specificity, called "Australia antigen/ 7 hepatitis-associated antigen (HAA), and now hepatitis B antigen (HB-Ag), has led to a great expansion in studies on the incidence and pathogenesis of hepatitis B, but the actual virions have not yet been unequivocally demonstrated (see Chapter 3). Serologically unrelated particles of similar morphology have been reported to occur in feces from patients with hepatitis A (Cross etal, 1971 ). Four diseases of similar nature, scrapie of sheep, transmissible encephalopathy of mink, and kuru and Creutzfeld-Jakob disease in man appear to be caused by similar agents, which differ from all known viruses by being nonimmunogenic. The causative agents are filtrable, highly heat-resistant, and highly resistant to ionizing radiation. It has been suggested that they may be small molecules of naked RNA, protected by being closely associated with cellular membranes (Diener, 1972a), but a definitive description of these agents is still awaited. This virus, which occurs as an inapparent infection in many laboratory mice and as a contaminant of cells and viruses derived from or passaged through mice (review: Notkins, 1965), shows some resemblances to the togaviruses. It appears to have an isometric core and a lipoprotein envelope, and its RNA is infectious. However, the viral RNA is large, perhaps 5 million dal tons (Darnell and Plagemann, 1972). In Germany, in 1967, a small outbreak of a serious new disease occurred in laboratory workers who had handled the tissues of recently imported vervet monkeys (review: Siegert, 1972). The causative agent grows in cultured cells and kills guinea pigs. Studies with inhibitors suggest that it contains RNA; of known viruses it most closely resembles rhabdoviruses in structure but is much larger and more pleomorphic (Murphy et ah, 1971b) . The foregoing account has shown how varied are the agents that we classify as viruses, for reasons based on their composition and their mode of intracellular replication. We can only speculate about their origins and relationships to each other, except in cases where the relationship is very close. It seems likely that different viruses belonging to any one genus, and in at least some cases, different genera allocated to a particular family, may be phylogenetically related. No use-34 1. Nature and Classification of Animal Viruses fui suggestions can be made concerning the relationships between families or genera (except in some of the cases where genera have been allocated to the same family), a fact which underlines the undesirability at this stage of our knowledge of erecting any taxa at levels higher than the family.Two suggestions have been made concerning the origin of viruses: (a) that they are the result of progressive parasitic degeneration of microorganisms (Green, 1935) and (b) that they have developed from components of the cells of their hosts (Andrewes, 1966; Luria and Darnell, 1967) , or are indeed still a permanent part of the host's genome (Todaro and Huebner, 1972) . With our present knowledge of the morphological and chemical complexity of the poxviruses, it is not difficult to envisage these agents as being the next degenerate step in the series: bacterium, rickettsia, chlamydia. Although they resemble bacteria in most important respects, rickettsia and chlamydiae are, like viruses, obligate intracellular parasites lacking the metabolic equipment for independent multiplication.On the other hand, some DNA viruses could well have arisen from episomes, by the acquisition of genetic information specifying a protein coat. Even this may not be essential, if Diener's (1972b) observations on potato spindle tuber virus are confirmed and generalized. The two alternatives are not mutually exclusive; some viruses may have evolved from cellular organelles like chloroplasts or mitochondria, themselves probably derived from bacteria (Swift and Wolstenholme, 1969). It is difficult to see where most RNA viruses could have originated except from cellular RNA's.Comparing the nearest neighbor nucleotide doublet frequencies of the nucleic acids of several large and small viruses, Subak-Sharpe (1969) noted that the patterns shown by small viruses with genomes of less than 5 million daltons (two enteroviruses, three parvoviruses, two polyoma viruses, and two papilloma viruses), closely resembled the pattern of mammalian DNA. On the other hand, the doublet frequency patterns of several viruses with large genomes (two herpesviruses and a poxvirus) differed strikingly from that of mammalian DNA. The doublet patterns of three adenoviruses resembled each other and showed a slight resemblance to the pattern of mammalian DNA, which could derive from some earlier natural fusion of genomes, like that recognized as a laboratory artifact with adenoviruses and SV40 (see Chapter 7). This evidence supports the notion that the small viruses may have originated from vertebrate cells whereas the herpesviruses, poxviruses, and probably the adenoviruses did not. These large viruses may have originated from the nucleic acid of cells of a different phylum, or as suggested earlier, by parasitic degeneration of microorganisms.