key: cord-0042458-06pj7zkp
authors: Kew, Olen M.; Nottay, Baldev K.; Obijeski, John F.
title: Applications of Oligonucleotide Fingerprinting to the Identification of Viruses
date: 2014-06-27
journal: nan
DOI: 10.1016/b978-0-12-470208-0.50008-6
sha: 8c2d674f728242e3f5ddbb1860c4e4f17cf512fc
doc_id: 42458
cord_uid: 06pj7zkp

This chapter focuses on applications of oligonucleotide fingerprinting to the identification of viruses. Fingerprinting is a technique by which oligonucleotides, produced by cleavage of RNA molecules with specific ribonucleases, are separated in two dimensions. It is a definitive method of identifying RNA viruses according to their genotypes. It is not subject to the problems of antigenic drift or antigenic convergence that complicate serological identification. Furthermore, it provides a semiquantitative means of following the evolution of viral genomes in nature. Because all regions of the genome are represented by the large diagnostic oligonucleotides, a survey of the total genomic changes can be monitored. Fingerprinting has two limitations as a diagnostic tool. First, although highly definitive, fingerprinting is not as rapid or inexpensive as serological techniques and cannot be as easily scaled up for routine identification of a large number of samples. Second, the evolutionary range of fingerprinting is short and relationships may not be evident for isolates of rapidly evolving viruses obtained over long intervals. However, these limitations are not large, compared to the full benefits offered to the virologist by the fingerprinting method.

Fingerprinting is a technique by which oligonucleotides, produced by cleavage of RNA molecules with specific ribonucleases, are separated in two dimensions. Originally developed as a preparatory method, fingerprinting is currently most frequently applied as an analytical method for comparing the genomes of RNA viruses. Comparisons are based on the principle that the large, structurally unique oligonucleotides separate into patterns, or "fingerprints," that are highly characteristic of the RNA sequence from which they derive. Accordingly, the sequence, and thus genetic, relation ships can be readily assessed. Since the characteristic oligonucleotides orig inate from all regions of the RNA molecule, the distributions of sequence similarities and differences may be surveyed over the entire viral genome. Although not providing specific base sequence information, fingerprinting does offer a simple and comparatively rapid means of sequence compari son, with the results obtained in a pictorial form.

The applications of fingerprint analysis in virology are very broad. A few examples of the applications include analysis of RNA genome size and com plexities, identification of subgenomic RNA molecules, analysis of the mes senger RNA species produced by both DNA and RNA viruses, characterization of defective-interfering viral genomes, biochemical de scription of genetic reassortants and recombinants, and the estimation of the degree of sequence divergence between closely related virus strains.

Oligonucleotide fingerprinting has emerged as an important tool for the identification of RNA virus strains (Brown, 1982; Clewley and Bishop, 1982) . Fingerprinting provides a reliable means of following the natural distribution and transmission of a specific viral genotype over time. This capability is of considerable value in human and veterinary epide miology, in which genetically distinct but antigenically similar epidemic and zoonotic strains in coetaneous circulation can be distinguished, and their transmission separately followed. Because viral antigens tend to be more conserved than the overall genomic RNA sequence, fingerprinting is gen erally much more sensitive than serology in detecting divergence between strains. On the other hand, fingerprint analysis is not complicated by antigenic drift. Although the availability of monoclonal antibodies promises to greatly improve the precision and discrimination of viral serodiagnosis (Yewdell and Gerhard, 1981) , fingerprinting is, in a sense, a more open technique, not requiring the production of new reagents in order to monitor the circulation of a given virus genotype. Further, unlike serology, finger print comparisons are not restricted to the small set of genomic sequences specifying the amino acid residues on a polypeptide surface.

In this chapter, we describe the principles and methodologies of oligo nucleotide fingerprinting and briefly review the applications of this tech nique to the identification of RNA viruses. Particular emphasis will be placed on the use of this method to follow the distribution and transmission of viral agents of infectious disease. Several excellent reviews on oligonu cleotide fingerprinting have appeared. Those of Beemon (1978) and Wang (1978) concentrate on the retrovirus group. Clewley and Bishop (1982) also describe methodology and the broad applications of fingerprinting in vi rology. Pedersen and Haseltine (1980b) and De Wächter and Fiers (1982) describe some current fingerprinting methods.

A. OVERVIEW OF FINGERPRINT ANALYSIS Oligonucleotides for fingerprinting studies are produced by complete digestion of viral RNA with the specific ribonucleases (RNases) Tl and A. RNase Tl, from the fungus Aspergillus, specifically cleaves phosphodiester bonds adjacent to guanosine (G) residues, producing guanosine 3'-phos phate (Gp) and a set of oligonucleotides of varying length terminating with Gp. Because of its high specificity under moderate digestion conditions, RNase Tl is the enzyme of choice for routine fingerprint studies. The other known single base-specific RNases also cleave at G (Takahashi and Moore, 1982) and thus confer no significant advantage over RNase T l . Pancreatic RNase A, which preferentially cleaves at pyrimidine residues, is infre quently used in routine comparative studies because of the reduced com plexity of the fingerprint patterns of its cleavage products. RNase A was used in conjunction with RNase Tl in classical studies of RNA structure (Barrell, 1971) , and has been applied more recently in detailed fingerprint comparisons between strains (Nomoto et al., 1981) as well as in studies on total genome sequences (Kitamura and Wimmer, 1980; Kitamura et al., 1981) . Many other RNases with different base specificities are known. Some are used in RNA sequence analysis to produce specific partial fragments of end-labeled RNA molecules (Simoncsits et al., 1977; Donis-Keller, 1980; Donis-Keller et al., 1977) . However, fingerprinting requires complete diges tion of the RNA, conditions for which the specificities of the other known RNases are too broad for them to be applicable.

The number of RNase Tl oligonucleotides in a complete digest is one more than the number of G residues in the RNA. For a random polynucleotide, the size distribution of oligonucleotide fragments is predicted from its composition by the relationship (Beemon, 1978) F n = ( G ) 2 ( l -G r -> where F n is the fraction of oligonucleotides of chain length n, and G is the mole fraction of guanosine residues in the RNA. For an RNA of 8000 bases containing 25 % G, this relationship predicts approximately 500 Gp mon omers (fl = l), 160 5-mers, 38 10-mers, 9 15-mers, 2 20-mers, and 6 oligon ucleotides for which A2>20, to occur among the 2001 oligonucleotides produced by complete digestion. Oligonucleotides with chain lengths n> 10 have a high statistical probability of containing unique sequences that occur only once in a haploid viral genome. For the hypothetical 8000-base viral genome described in the example above, oligonucleotides containing 10 or more base residues constitute approximately 25% of the total sequences. Although the structurally unique oligonucleotides of a typical RNA mole cule are too numerous to be resolved effectively by a single mechanism of separation, the two-dimensional fingerprinting technique, in which a dif ferent separation mechanism is employed for each dimension, allows good resolution to be achieved in practice. The proportion of the genome that can be resolved experimentally into separate oligonucleotide spots is pri marily dependent upon the size and complexity of the RNA. This is be cause, for polynucleotides of a given guanosine content, the number of unique oligonucleotides increases with chain length, but the mean oligo nucleotide chain length remains constant (Aaronson et al., 1982) . For large RNA molecules, such as viral genomes, an important fraction of the po tentially informational oligonucleotides is incompletely resolved, and is thus effectively unavailable for analysis. For nonsegmented genomes, from 5 to 15% of the unique oligonucleotides can be resolved and individually ana lyzed. With viruses having segmented genomes it is possible to examine higher percentages of the total genome by separate analysis of each seg ment.

Two different fingerprinting methods have been developed. The earlier method, developed by Brownlee and Sanger (1969) , utilized high-voltage electrophoresis in the first dimension and chromatography in the second. In the second method, that of De Wächter and Fiers (1972) , separation is achieved in both dimensions by electrophoresis through polyacrylamide gels. In both methods, oligonucleotides are separated in the first dimension ac cording to their net charges and in the second dimension primarily accord ing to chain length.

First-dimension electrophoresis for both is generally performed near pH 3.5 in the presence of 6-7 M urea. At this pH, the greatest charge differ ences among the bases occur (Markham and Smith, 1952) , such that the net negative charges, and thus electrophoretic mobilities, of the nucleoside monophosphates increase in the order C p < A p < G p < U p . In the absence of significant frictional forces,'oligonucleotides migrate according to the total charge contributions of their constituent base residues. The combi nation of acidic pH and high urea concentrations disrupts the secondary structure of the oligonucleotides and reduces interchain aggregation, largely excluding effects of conformation on mobilities. Further, for the polyacrylamide gel system, the use of large-pore gels eliminates molecular siev ing effects on all but the largest oHgonucleotides (Frisby et al., 1976) , thus maximizing the contributions of charge to electrophoretic mobilities.

Second-dimension separation in the two fingerprinting systems occurs by fundamentally different mechanisms. The method of Brownlee and Sanger (1969) utilizes homochromatography, in which oHgonucleotides are com petitively displaced from a cationic surface during ascending chromatography. The displacement rate, and thus migration rate, is approximately inversely related to chain length, but base composition and secondary struc ture significantly influence mobilities. De Wächter and Fiers (1972) em ployed a highly cross-linked 20% polyacrylamide gel buffered at pH 8 ("neutral gel"). In the neutral pH range, the four bases are largely un charged. All significant charges on the oHgonucleotides therefore derive from ionization at the phosphodiester bonds (and at the terminal phos phate), and the charge:mass ratios for all long oHgonucleotides are nearly constant. Thus the polyacrylamide gel acts strictly as a molecular sieve per mitting fractionation according to size and shape. To a first approximation, electrophoretic mobilities in the second dimension decrease with the loga rithm of the oligonucleotide chain length (De Wächter and Fiers, 1982) . Deviations from this relationship, arising from contributions of secondary structure and base composition, are usually small (De Wächter and Fiers, 1982; Lee and Wimmer, 1976; Pedersen and Haseltine, 1980a) .

A typical two-dimensional polyacrylamide gel fingerprint of a virus RNA is shown in Fig. 1 . In this example, the smaller oHgonucleotides have been run off the second-dimension gel in order to improve resolution. The oH gonucleotides are arranged in a graticulated pattern formed by a series of bands curving diagonally away from the quadrant nearest the origin of firstdimension electrophoresis (lower left, Fig. 1 ). This pattern is most evident in the region of the more numerous shorter-chain oHgonucleotides. The bands form according to uridine (U) content. OHgonucleotides of the band closest to the origin contain no U residues, those of the next band contain one U, those of the next, two, and so on. Within each band, the cytidine (C)-rich oHgonucleotides are diagonally opposite (above and behind in the orientation of the Fig. 1 pattern) the adenosine (A)-rich oHgonucleotides of equivalent chain length (Pedersen and Haseltine, 1980a; De Wächter and Fiers, 1982) . The lower part of each fingerprint contains the large, RNase Tl-resistant oHgonucleotides, whose mobilities are diagnostic for a virus genotype. Fingerprints of related and unrelated polioviruses are shown in Fig. 2 . The diagnostic oHgonucleotides are distributed in a roughly trian gular region, a pattern that occurs because with increasing chain length oH gonucleotides generally approach the mean base composition, and hence first-dimension mobility, of the entire RNA molecule.

-► 2nd DIMENSION Note that the vaccine-related isolates (center column) have fingerprints that are very similar to those of the respective oral vaccine reference strains (left column). Wild isolates from each of the three serotypes have fingerprints (right column) that are entirely distinct from those of the vaccinerelated group, indicating a lack of close genetic relatedness. Note also the strong similarities in the fingerprints of Sabin type 3 (Leon 12 a,b) and its nonattenuated parent, Leon (Fig. 1 Oligonucleotide fingerprinting is a tool of extraordinary power and pre cision for demonstrating close genetic relationships among viruses and, therefore, for determining the natural distribution and transmission of any given viral genotype. The typical fingerprint contains 30-60 diagnostic spots, each having an independent mobility and, as a result, they represent many independent points of reference for comparison. Because the potential number of patterns formed by the spots of the diagnostic oligonucleotides is very large, equivalent migration of a majority of the spots of two viral RNA samples implies a close genetic relationship between the viruses. Fre quently fingerprints of different isolates are compared by simple visual in spection, and similarities in patterns can be perceived when as few as 50% of the diagnostic oligonucleotides spots are shared. The simplest method for confirming genetic relatedness is codigestion and co-electrophoresis of equivalent isotopie amounts [as measured in disintegrations per minute (dpm)] from the two viral RNAs. This approach appears to be sufficient when 50% or more of the oligonucleotide spots comigrate. When fewer spots are shared, additional evidence, such as composition or sequence analysis of corresponding spots, should be obtained to establish genetic re latedness clearly. Moreover, when less than about 25% of the oligonucleo tide spots of two different strains comigrate, the remaining pattern similarities are usually difficult to detect, and further comparisons may not be performed unless genetic relatedness is suspected on other grounds, such as epidemiological data or the existence of strains with intermediate fin gerprint patterns.

The number of oligonucleotide spots common to two RNA molecules decreases rapidly with differences in their overall sequence homology. For example, Aaronson et al. (1982) used computer-simulated mutation of the genome segment encoding the influenza hemagglutinin to examine the re lationship between the number of large oligonucleotides shared and the ex tent of overall base sequence homology. According to their model, two RNA molecules that diverge at 1, 5, or 10% of their total sequences share 85, 50, and 25%, respectively, of their diagnostic oligonucleotides. These estimates are in good agreement with experimental comparisons of foot-and-mouth disease virus genomes by fingerprinting and competition hybridization (Robson et al., 1979) . From these studies it is clear that the lower threshold for recognition of relatedness by fingerprinting is near the level of 90% base sequence homology. Thus, the evolutionary range of fingerprinting is quite short.

The basis for the pronounced sensitivity of oligonucleotide fingerprinting to genetic divergence is evident in view of the mechanisms of fractionation.

Consider a typical diagnostic oligonucleotide of 15 nucleotide residues. A change in only one base (7% of the total) involving C, A, or U residues will result in an altered first-dimension mobility, and the spot will appear to have shifted laterally in the fingerprint of the mutant RNA. The greatest mobility shifts are expected from C->U transitions, the smallest from C-> A trans versions. Changes involving G residues will either eliminate or create a cleavage site for RNase T l , and spots may appear to move both laterally and vertically within the diagnostic region, or disappear from it altogether. Similarly, new diagnostic spots may arise from other regions of the genome upon mutation of a G residue.

With the exception of genome sequencing, other approaches for mea suring genetic relatedness, both molecular and serological, are less sensitive than fingerprinting in detecting small mutational changes, but, instead are able to detect more distant relationships among viruses. For example, com petition nucleic acid hybridization can detect homologies between strains that share only 25% of their overall sequences (Minson and Darby, 1982) . Restriction endonuclease analysis of DNA genomes may be used for quan titative estimates of sequence divergence as great as 20-30% . Serological methods, especially those involving polyclonal antibodies directed against internal or nonstructural viral protein antigens, are often able to detect conserved homologies in polypeptide structure that are be yond the limits of sensitivity of standard nucleic acid hybridization meth ods. Genome sequencing has the ultimate sensitivity to detecting mutational differences among viruses, and routine comparative sequence analysis of selected genome regions is possible. However, unlike fingerprinting, current sequencing methods are not well suited for providing a survey of the total sequence divergence among many strains.

A. PREPARATION OF RADIOLABELED RNA

The genomic RNAs of most animal RNA viruses can be labeled to ac tivities suitable for production of high-quality fingerprints by biosynthetic incorporation of ortho[ 32 P]phosphate. The general strategies for isotopie labeling of viral macromolecules have been reviewed by Henry (1967) . Con-ditions for optimal incorporation of radiophosphate vary for different virus and cell culture systems. Important factors to be considered first when de veloping a labeling strategy are duration of the replicative cycle and the multiplicity of infection of input virus.

For viruses having short infectious cycles and for which high-titer inocula are available, infection occurs as a one-step cycle. Labeling periods may precede or coincide with infection and extend throughout the cycle, or be adjusted to correspond to an optimal interval within the cycle. Efficiencies of incorporation may be improved by maintenance of the cells in phos phate-deficient medium for several hours prior to infection, thus depleting intracellular pools of inorganic phosphate. Because exogenous phosphate equilibrates slowly with intracellular nucleotide pools (Henry, 1967) , label should be added well before the expected period of maximal RNA synthesis. Chilling (0-4°C, 60-120 min) infected suspension cultures immediately prior to addition of 32 P has been used to improve incorporation of label into poliovirus RNA (Wimmer, 1972) . For high-specific-activity labeling, ortho[ 32 P]phosphate is added without carrier to a synthetic medium lacking phosphate. If serum is to be included in the medium, it should be exten sively dialyzed against 0.15 M NaCl. The volumes of the labeling medium should be kept to a minimum, and frequent agitation of the medium is recommended for uniform distribution of isotope. The increased cell den sities (1-5 x 10 6 cells/ml medium) generally used during radiolabeling may necessitate inclusion of supplemental buffers, such as 25 mM Af-2-hydroxyethylpiperazine-W-2-ethanesulfonic acid (HEPES) or 0.3% sodium cit rate, pH 7.2 (Ogra et al., 1968) . These buffers are occasionally inhibitory, and their effects on total yield should be determined before routine use.

For systems in which virus production occurs over an extended period or that require multiple rounds of infections, optimal labeling conditions should balance total virus yield against the specific activity of the final RNA product. Cell viability is prolonged by inclusion of some carrier phosphate in the culture medium, and multiple harvests may be required to assure recovery of a high proportion of intact virions. In all systems, virus should be harvested and purified as early as possible to minimize radiation damage from unincorporated isotope.

Typical inputs of ortho[ 32 P]phosphate are 100-1000 /xCi/ml culture. The total isotope input should be adjusted according to expected yield of label incorporated into intact viral RNA, which may vary even among closely related virus strains, from nearly 1% to less than 0.01%. In some systems, inclusion of actinomycin D (1-5 /xg/ml) in the labeling medium may in crease incorporation of 32 P into virus RNA by selectively inhibiting host transcription. However, the effects of this antibiotic on labeling efficiencies should be examined carefully before routine use, especially when many dif-ferent agents are to be compared. For example, some strains of poliovirus, a group generally regarded as actinomycin D resistant, incorporate 32 P poorly in the presence of this antibiotic (Nomoto et al., 1979a; .

The highest quality fingerprints are obtained with RNA labeled in vivo and extracted from purified virions. Some rapid diagnostic fingerprinting procedures utilize labeled intracellular viral RNA (La Torre et ai, 1982) . RNA labeled in vivo is frequently described as "uniformly labeled," im plying that each phosphodiester bond is labeled to an equivalent specific activity. Uniformity of label is demonstrated by equivalence of base com position whether the RNA is degraded to 3 ' -nucleotides (by RNase T2 digestion) or to 5 '-nucleotides (by digestion with nuclease PI or snake venom exonuclease). Although demonstrated in some systems (Billeter et al., 1974) , deviations from uniformity have been observed in others, com plicating determinations of oligonucleotide base composition (Lee et al., 1979) . With long-chain oligonucleotides, the specific activity differences among residues tend to average out, so that the radioactivity in a large oligonucleotide labeled in vivo is in close proportion to its chain length.

In vitro labeling of viral RNA presents several advantages over labeling in vivo. For some viruses, in vitro methods offer the only effective means of labeling genomic RNA to high specific activities (Frisby, 1977) . More over, in vitro isotopie labeling can be performed rapidly and requires sub stantially lower inputs of radioisotope. Only small quantities (1 /xg or less) of chemically pure viral RNA are needed for analysis. Repeated in vivo labelings are avoided, as it is generally necessary to prepare a viral RNA sample only once, with small amounts being withdrawn for labeling when ever needed. Finally, 5'-and 3'-radiophosphate end-labeled oligonucleo tides can be recovered from the fingerprints for sequence analysis.

Three basic approaches are applied for in vitro labeling of RNA for fin gerprinting.

a. 5 '-End Labeling. The most widely used in vitro labeling procedure utilizes T4 polynucleotide kinase (PNK) to transfer radiophosphate from [γ-32 Ρ]ΑΤΡ to the free 5'-hydroxyl groups of RNase-generated oligonu cleotides (Frisby, 1977) . Alkaline phosphatase is added to the RNase Tl digestion mixture to remove all 3 ' -terminal phosphates from the oligonu cleotides. The combined activities of alkaline phosphatase and PNK cata lyze a net transfer of phosphate from the 3 ' -to the 5 ' -ends, a process that does not significantly alter the electrophoretic mobilities of the treated oli gonucleotides (Brownlee and Sanger, 1967) . Also, because some PNK prep arations contain a contaminating 3 '-phosphatase activity (Szekely and Sanger, 1969) , prior removal of the 3'-terminal phosphates eliminates po tential charge and migration heterogeneity brought about by partial dephosphorylation of an oligonucleotide. Several approaches have been used to eliminate alkaline phosphatase activity from the subsequent PNK reac tion. These include removal by phenol extraction (Frisby, 1977) , inactivation with 0.1 TV H N 0 3 (Lee and Fowlks, 1982) , or inhibition by inclusion of inorganic phosphate in the PNK reaction (Chaconas and van de Sande, 1980) . The last approach, included in the labeling method of Pedersen and Haseltine (1980b) , is operationally very simple. We have used their method, described below, to obtain high-quality fingerprints of viral RNA.

Dried RNA (200-500 ng) in a 0.5-ml polypropylene tube (Eppendorf) is resuspended in 1 μΐ of sterile water to which is added 1 μΐ of 40 mM Tris-HC1 (pH 8.0) containing for each microgram of RNA 0.2 unit of RNase Tl (Sankyo, distributed by Calbiochem) and 2 x 10" 4 units of Escherichia coli alkaline phosphatase (BAP) (Worthington BAPF; P/L Biochemicals). It is important to omit EDTA, normally present during RNase Tl digestion of prelabeled RNA, because it is a strong inhibitor of BAP. For less than 100 ng of RNA, addition of 100 ng of polyguanylic acid (Miles Biochem icals) to the digests facilitates cleavage by RNase Tl (Pedersen and Has eltine, 1980b) . Incubation is for 30 min at 37°C. The PNK reaction immediately follows by addition of 50 μΐ of PNK reaction mixture and 5 units of PNK (P/L Biochemicals; New England Nuclear). The PNK reac tion mixture is prepared by first lyophilizing 50-200 μ€\ [γ-32 Ρ]ΑΤΡ in a 0.5-ml Eppendorf tube, then dissolving the labeled ATP (>2000 Ci/mmol) in 50 μΐ of 10 mM K 2 HP0 4 -K 3 P0 4 (pH 9.5), 10 mM Mg(OAc) 2 , 5 mM dithiothreitol. Incubation is at 37°C for 1-16 hr. Because the rates of phosphorylation by PNK vary widely for different oligonucleotides (Frisby, 1977) , prolonged incubation favors more uniform incorporation of label into the oligonucleotides. However, with increased incubation, nuclease contamination of the PNK preparations may become evident, and the op timal conditions may involve a compromise between maximizing uniform incorporation and minimizing the appearance of secondary spots. The re action is terminated by addition of 50 μΐ of 0.6 M NH 4 OAc, and 100 μg of carrier yeast tRNA. The oligonucleotides and carrier are precipitated by mixing with 300 μΐ of cold (-20°C) 95% ethanol and chilling in a dry iceethanol bath for 20 min. The precipitate is collected by centrifugation in a microcentrifuge (15 min at 4°C), washed with 400 μΐ of cold 95% ethanol, and centrifuged again (5 min at 4°C). After lyophilization, the labeled di gest is ready to be resuspended in first dimension electrophoresis sample buffer.

b. 3 '-End Labeling. Use of the base-specific chemical cleavage method of Peattie (1979) for sequence determination of oligonucleotides requires radiolabeling of the 3'-ends. Currently, this is most frequently accom-pushed by use of the T4 RNA ligase-mediated transfer of cytidine 3 ' -[5'-32 P]diphosphate (pCp) to the 3'-terminal hydroxyls of dephosphorylated oligonucleotides (England and Uhlenbeck, 1978) . Lee and Fowlks (1982) have combined this method with fingerprinting to produce 3 ' -labeled oligonucleotides by a simple procedure. Only an outline of the procedures is given here, as detailed descriptions are presented in the original articles. Digestion with RNase Tl and BAP is performed essentially as described above in the subsection on 5'-end labeling. Afterward, BAP is inactivated by making the digest mixture 100 mM in H N 0 3 (pH 2; 10 min at room temperature), followed by addition of ZnS0 4 to a final concentration of 1 mM (2 min at room temperature) to inactivate RNase T l . The digest is precipitated with ethanol, washed with 95% ethanol, and lyophilized. The precipitate is resuspended in the ligase reaction system (30 μ\) containing 50 mMHEPES (pH 8.3), 5 mM ATP, 10 mMMgCl 2 , 3.3 mM dithiothreitol, 10% (v/v) dimethyl sulfoxide, 15% (v/v) glycerol, 200 μ€Ί [5'-32 P]pCp, and 6 units of T4 RNA ligase (P/L Biochemicals). Incubation (4°C, 16 hr) is terminated by ethanol precipitation of the labeled oligonucleotides as de scribed for 5'-end labeling.

c. Radioiodination. Commerford (1971 Commerford ( , 1980 ) described a method for the radioiodination of cytosine residues of RNA in the presence of thallium ions that is applicable to oligonucleotide fingerprinting (Robertson et al., 1980; Clewley and Bishop, 1982; Lee and Fowlks, 1982) . Radiolabeling with 125 I has the advantages of being very rapid, producing labeled RNA with high specific activities and having a longer half-life (60 days) than 32 P-labeled RNA (14.3 days). The factors governing the efficiencies of iodine labeling are discussed in detail elsewhere (Huang and Pagano, 1977; Com merford, 1980) . We have used the method of Lee and Fowlks (1982) , de scribed below, to produce fingerprints of high quality.

From 1 to 20 μg of viral RNA in water (5 μ\ in a 1.5-ml Eppendorf tube) is combined with 12.5 μ\ of 0.3 mM thallic nitrate buffer: 30 μ\ of 0.1 M thallic nitrate [Tl(NO) 3 ; K and K; Alfa] in 1 TV H N 0 3 mixed with 9.97 ml of 0.2MNaOAc (pH 4.4). Carrier-free 125 I (0.3-1.5 mCi, pH 7-11) is added, and the total volume is brought to 25 μΐ with water. After 4 min of incu bation at 70°C, 475 μ\ of sodium dodecyl sulfate (SDS) buffer [10 mMTris-HC1 (pH 7.5), 0.12MNaCl, 0.1% SDS, 1 mMEDTA] is added, and the mixture is incubated for an additional 10 min at 70°C. Then 1 ml of cold 95% ethanol is added, and the mixture is chilled in dry ice-ethanol for 20 min, then centrifuged (30 min, 4°C). The precipitate is washed four or five times by gently filling the tube with cold 80% ethanol and decanting. The final supernatants are monitored with a Geiger counter; after confirmation that no soluble label remains, the pellet is lyophilized. The labeled RNA is ready for digestion with RNase T l .

d. Comments. Fingerprints produced from oligonucleotides labeled in vitro generally correspond well with fingerprints of the same RNA labeled in vivo. However, depending on the method of labeling, small differences may exist. As mentioned, spots of 5'-or 3'-end-labeled oligonucleotides may vary in intensity as a result of preferential activities of PNK and RNA ligase (Frisby, 1977; Lee and Fowlks, 1982) . Moreover, the total number of spots may differ among fingerprints labeled in vivo and in vitro. Spots of 5'-labeled material, for example, may be missing if PNK phosphorylation is very inefficient or blocked by the presence of a 5 ' -terminal cap struc ture (Banerjee, 1980) or a protein (Wimmer, 1982) . Spots may also be missing from fingerprints of 3'-labeled material (Lee and Fowlks, 1982) , possibly as a result of incomplete dephosphorylation. Secondary oligonucleotide fragments, produced by trace nucleases, may also be labeled, thus giving rise to additional spots (Szekely and Sanger, 1969) . Fingerprints of iodinated RNA may also be deficient in certain spots. Because radioiodination of RNA occurs primarily by the production of 5iodocytosine, 5-iodouracil being a minor product (Commerford, 1971) , only cytosine-containing oligonucleotides are effectively labeled. Although the larger diagnostic oligonucleotides generally contain some cytosine, the spots of cytosine-deficient oligonucleotides may be light. The poly (A) streak ob served in the fingerprints of in v/vo-labeled RNA of many positive stranded viruses is absent from fingerprints of iodinated RNA.

Fingerprints produced with in v/Yro-labeled material are generally satis factory for comparisons of the genomic RNAs of different viruses. Only when the fingerprints of the most closely related strains are being compared does the potential existence of artifactual spots present uncertainties in es timating the extent of sequence divergence.

Digestion of in v/vo-labeled RNA or iodinated RNA is performed in 1.5ml Eppendorf tubes. Viral RNA plus 100 μg of carrier yeast tRNA are dis solved in 1 ml of 20 mMTris-HCl (pH 7.8), 150 mMNaCl, transferred to a 1.5-ml Eppendorf tube, and mixed with 200 μΐ of 1% (w/v) cetyltrimethylammonium bromide (CTAB; obtained from Sigma or BDH). The CTAB precipitation step purifies the RNA of uncharged polymeric contam inants and nucleases that are precipitated by ethanol (Ralph and Bergquist, 1967) . After 20 min on ice, the insoluble cetyltrimethylammonium ribonucleate salts are collected by centrifugation in a microcentrifuge (5 min at 4°C). The flocculent pellet is washed twice with cold 70% ethanol/30% 150 mM NaCl in water, to convert the RNA back to the sodium salt, then once with 70% ethanol-water, and finally with 95% ethanol. Each wash step is followed by 2 min centrifugation at 4°C. The RNA is dried under vacuum, and the pellet is resuspended in 20 μί of 10 mM Tris-HCl (pH 7.5), 1 mM EDTA, containing 0.2 unit of RNase Tl per /xg RNA (enzyme: RNA mass ratio of 1:25). Samples of double-stranded RNA are denatured by boiling for 2 min and quick chilling (Walker et al., 1980) or freezing (Sugiyama et al., 1982) before addition of enzyme. Digestion is at 37°C for 30 min. To 20 μ\ of digest is added 40 μΐ of urea-dye mix (9 M urea/50% glycerol satu rated with bromphenol blue/0.3% trypan red).

Digestion of RNA with RNase A (Calbiochem) is at 37°C for 30 min in 10 mM Tris-HCl (pH 7.5), 1 mMEDTA, 0.3 MNaCl, at an enzyme:RNA mass ratio of 1:10. Increased ionic strength suppresses secondary cleavage by RNase A at adenosine residues. However, the salt will impair resolution in the first dimension and should be removed by ethanol precipitation prior to electrophoresis (Nomoto et al., 1981) .

Stocks of the RNase Tl and RNase A are stored as small aliquots in their respective digestion buffers at -20° C at concentrations of 1 mg/ml.

Many variations of the basic technique of De Wächter and Fiers (1972) have been described (Frisby et al., 1976; Lee and Wimmer, 1976; Kennedy, 1976; Clewley et al., 1977b; Pedersen and Haseltine, 1980a,b; Stewart and Crouch, 1981) . Our fingerprinting procedures essentially follow the modi fications described by Lee and Wimmer (1976) and Lee et al. (1979) .

a. Reagents. All stock solutions (Tables I and II) should be prepared with deionized water using chemical reagents of the highest grades avail able. Especially critical to the production of high quality fingerprints are the chemical purities of the acrylamide Ν,Ν' -méthylène bisacrylamide (Bis) and urea components.

Highly purified, electrophoresis-grade acrylamide and Bis from com mercial sources (e.g., Bio-Rad Laboratories; Eastman Organic Chemicals) can be used without further purification. Reagent-grade acrylamide can be purified by recrystallization in chloroform, and Bis by recrystallization in acetone, as described by Adesnik (1971) . Alternatively, monomers can be mixed in the specified ratios (Tables I and II ) and purified by stirring (12 hr, room temperature) with mixed-bed resin (20 g/liter; Bio-Rad AG 501-X8), followed by filtration through Whatman No. 1 paper. Purified solu tions are passed through a Whatman GF/C filter and stored in the dark at 4°C.

Solutions of ultrapure urea (Research Plus; BDH AnalAR) require only filtration (0.22 /xm) before use. Reagent grade urea is purified by stirring (12 hr) at 9 M solution with mixed-bed resin (20 g/liter). The urea solution is further purified by filtration through a bed of silicic acid (Lee and Wimmer, 1976) , prepared by suspending about 100 g of powdered silicic acid (Sigma SIL-R) in 1 liter of deionized H 2 0, and filtering the suspension un der suction through a 15-cm filter of Whatman No. 4 paper. Solutions of purified urea may be stored for up to 3 months at 4°C. Saturated citric acid is stored at room temperature. Hydrogen peroxide (30%) is dispensed into 1-ml volumes and stored at 4°C.

Electrophoresis grade A^A^TV^TV'-tetramethylethylenediamine (TEMED; Bio-Rad Laboratories) is stored at 4°C.

Solutions of ammonium persulfate (10%) are prepared fresh before use. Solutions of Tris-borate (1.0 M; 121 g/liter Tris base and 63 g/liter boric acid adjusted to pH 8.2) are autoclaved and stored at room temperature.

b. Apparatus. First-dimension plates (12 x 45 cm) are cut from 1/8-in. glass. Second-dimension plates are made from 3/16 in. glass cut to 36 x 43 cm. Spacers are 1/16 in. (0.16 cm) x 1 cm strips of polyvinyl chloride, cut to lengths of 45 cm for the first-dimension assembly and 43 cm for the second-dimension plates. The first-dimension well former is 1 /16 in. Teflon with two or three teeth, 1.25 cm wide x 4 cm long, separated by 1.25 cm. Plates should be cleaned immediately after use with a mild nonabrasive detergent, rinsed thoroughly with deionized water, and dried. Just before use, plates are washed with 95% ethanol and air dried. Cleaning acid should be avoided, because the glass surface may be altered, causing the first-di mension gels to adhere so tightly to the glass that removal and transfer are prevented.

The plates and spacers are assembled to form a mold for the gel, and the assembly is clamped with medium steel binder clips. Precautions must be taken to prevent leakage of the unpolymerized gel from the mold. We prefer to seal the sides and bottom with 1-in. Teflon tape (3M catalog No. 5490) taking special care to reinforce the corners, where leaks are most prone to occur. Alternatively, spacers may be coated with a thin film of silicone vac uum grease before assembly. Chamber bottoms may also be sealed by use of a temporary spacer (Clewley et al., 1977b) or with plasticine (De Wächter and Fiers, 1972) . Problems of leakage of the first-dimension gels are re duced by setting the chamber at a nearly horizontal angle during polymer ization. To assure that the first-dimension gel strip is in firm contact with the glass plates during polymerization of the second-dimension gel, the bot tom of the assembly is clamped with several steel binder clips, and air pock ets are removed by hydrostatic pressure. Good contact can be achieved also by using slightly thinner spacers for the second-dimension chamber than those used for the first dimension (Pedersen and Haseltine, 1980a) , or by using spacers of compressible silicone rubber (Clewley et al., 1977b) .

The electrophoresis units used in our laboratory, purchased from Lee's Unique Instruments (De Soto, TX), are shown in Fig. 3 . The slab gel cham bers are placed in the lower buffer reservoirs and connected to the upper reservoirs with filter paper wicks of Whatman 3 MM or equivalent, precut to 9 x 9 cm (first dimension) and 6.5 x 34 cm (second dimension). The wicks are saturated with electrode buffer, and the exposed parts are wrapped in Saran wrap or between Mylar sheets to prevent evaporation. Upper and lower reservoir capacities are 375 ml each for the first-dimension unit and 1000 ml each for the second-dimension apparatus. Although the seconddimension unit can accommodate up to three gels, we use only one gel per unit and do not recirculate reservoir buffers.

c. First-Dimension Electrophoresis. The first-dimension gel is 8% polyacrylamide (acrylamiderBis ratio, 80:1), 6 M urea, pH 3.3. The electrode buffer is 25 mM citric acid adjusted to pH 3.3 with 10 N NaOH. Stock solutions and mixing ratios are given in Table I , and the conditions for electrophoresis are shown schematically in Fig. 4 .

The acrylamide-urea solution is titrated to pH 3.3 with saturated citric acid (usually less than 1 ml is required; substantially greater amounts in dicate the presence of impurities in the urea). The solution is warmed to 37°C, then mixed with the catalysts of Jordan and Raymond (1969) in the order FeS0 4 *7H 2 0, ascorbic acid, and hydrogen peroxide, and poured to within about 1.5 cm from the top of the prewarmed (37°C) mold. A 9 X 10 cm sheet of Parafilm, inserted about 1 cm into the mold and supported and curved with one hand, serves as a convenient pouring guide. Bubbles are removed by briefly holding the mold upright. The mold is then set at an angle of about 15°, and the well former is inserted about 2 cm into the gel.

After polymerization (30 min), the bottom tape seal and the well former are removed, and the top of the gel is washed with and then overlay ed with electrode buffer. The sample (up to 60 μΐ and 100 μg of RNA digest), in loading buffer, is briefly heated (60°C for 3 min) to disrupt aggregates, and layered in the sample well with a microsyringe (Hamilton). The loaded gel is transferred to the electrophoresis unit in a cold room, and the wick is inserted, saturated with electrode buffer, and wrapped. Electrophoresis is performed at 4°C at a constant potential of 450 V (about 7 mA). Total power (watts = volts x amperes), a measure of heat production, should not exceed 3.2 W. The run is terminated when the bromphenol blue dye marker (green at pH 3.3) migrates 21.5 cm from the bottom of the sample well (about 16 hr). The trypan red marker migrates ahead of the bromphenol blue (Fig. 4) , but behind the leading oligonucleotide fragments. d. Second-Dimension Electrophoresis. After first-dimension electro phoresis, a glass plate is removed, and, using a Plexiglas template as a guide, the gel is cut with a rotary pizza cutter into 2.5 cm-wide strips containing the migration lanes of oligonucleotides. The strips are cut at their ends 19.5 cm above and 14 cm below the bromphenol blue dye mark (Fig. 2) . A piece of Parafilm, about 4 x 45 cm, is centered and then pressed gently on top of a gel strip, with care taken to remove air pockets. One end of the Para film and gel strip is gently lifted with forceps and pulled directly back to transfer the gel, without stretching, from the glass plate to the Parafilm strip. At this point the gel strips may be frozen at -70°C if desired. Oth erwise, each strip is soaked for 5 min in 500 ml of second-dimension elec trode buffer (50 mM Tris-borate, pH 8.2) to remove urea, the presence of which will otherwise inhibit polymerization at the junction between the firstand second-dimension gels. The lower reservoir of the second-dimension electrophoresis apparatus is a convenient vessel for this operation, after which the buffer is discarded. The strip is drained of buffer and placed, gel side up, along the edge of a laboratory bench. A second-dimension glass plate is placed over the gel, and contact is made when the strip is centered, parallel to, and about 1-2 mm from, the short (36 cm) edge of the plate. The ends of the Parafilm are folded over the plate to hold the gel in place, and then the plate is turned over, and the Parafilm is removed. After as sembling, sealing, and clamping the chamber, about 10 ml of second-di mension electrode buffer is dispensed into the chamber, and the chamber is placed upright in a tank of cold tap water (40 cm deep) to allow the hydrostatic pressure to express bubbles between the gel and the glass plate. The chamber is removed from the tank, and the buffer is allowed to overlay the gel for 10 min, then poured out.

The second-dimension gel is 22% polyacrylamide (acrylamide:Bis ratio, 15:1) in 50 mM Tris-borate, pH 8.2. Components and mixing volumes for the second-dimension gel are given in Table II . A 10% low-cross-linked polyacrylamide connecting gel is cast first to form a sealant and junction between the first-and second-dimension gels. After addition of catalysts, about 5 ml of connecting gel solution is pipetted down a spacer; the cham ber is quickly tilted from side to side to assure good coverage of the gel strip and then set on a level surface for polymerization (1-2 min). The sec ond-dimension separating gel solution is transferred (via tubing attached to a glass reservoir) until the chamber is about one-third full. The chamber is tilted as before to mix any remaining unpolymerized connecting gel, placed upright in the water tank, filled to about 2 cm from the top, and then ov erlaid with water. Immersion at this stage serves two purposes. The cold water acts as a heat sink to reduce convection during polymerization, and the balance in hydrostatic pressure retards leakage of material from the chamber (Lee and Wimmer, 1976) . Polymerization is complete in about 30-40 min.

Electrophoresis is performed at room temperature at a constant voltage of 600 V (about 16 mA/gel), or at a constant power of 10-12 W/gel. Reg ulation at constant power is preferred because as the resistance of the sec ond-dimension gel system increases during the run, the voltage may be allowed to rise (up to 1000 V) under conditions of constant Joule heating. For recovery of all size classes of oligonucleotides, electrophoresis is ter minated when the bromphenol blue dye migrates 20 cm (about 16 hr). Al though separation in the first dimension is limited by the size of the gel strip that can be accommodated by the second-dimension gel mold, reso lution of the diagnostic RNase Tl-resistant oligonucleotides is improved in the second dimension by continued electrophoresis. Migration of the brom phenol blue dye marker to 30 cm or more is frequently advantageous.

e. Comments. Several conditions must be carefully controlled to assure production of high-quality fingerprints. It is particularly important to avoid excessive heating of the gels during electrophoresis. Overheating in the firstdimension separation may cause extensive streaking of the oligonucleotides. Poor contact between the first-and second-dimension gels, caused, for ex ample, by incomplete removal of urea, may also result in localized heating and thus produce distorted fingerprints. Bubbles in the gels or nonuniform polymerization of the second-dimension gel can also cause distorted pat terns.

The formulations described here incorporate several important modifi cations of the original De Wächter and Fiers (1972) fingerprinting system. The Bis cross-linker content is lower in the first-dimension and higher in the second-dimension gels. Urea is omitted from the first-dimension elec trode buffer. First-dimension electrophoresis is run at pH 3.3 because it was found empirically that better separation of the oligonucleotides was obtained (Lee and Wimmer, 1976) . The original Tris-citrate second-dimen-sion buffer system is replaced by a Tris-borate system, of lower conduc tivity (Frisby et al., 1976) , thus allowing for higher voltage gradients. Additional changes have been introduced by other investigators. Polymer ization of the first-dimension gel may be catalyzed with TEMED and am monium persulfate (Holland et al, 1979) . Pedersen and Haseltine (1980b) omitted urea in the first-dimension gels and included 7 M urea in the sec ond-dimension gels. Shorter oligonucleotides are better resolved, and the urea-containing gels may be frozen without breaking at -70°C, allowing for higher fluorescent yields with intensifying screens (Laskey, 1980) . Sec ond-dimension gels may also be frozen if they contain 0.5% linear polyacrylamide (Freeman et al., 1979; Stern and Kennedy, 1980) .

The original systems for fingerprinting complete viral genomes were based on the homochromatographic technique of Brownlee and Sanger (1969) . Homochromatography was developed to overcome problems encountered earlier in resolving long-chain oligonucleotides (Sanger et al., 1965) , and its availability opened the way for a series of important structural and com parative studies of RNA viral genomes (Horst et al., 1972; Duesberg and Vogt, 1973; Billeter et al., 1974; Wangetal., 1975) . Homochromatography is also used in sequence analysis of end-labeled RNA (Rensing and Schoenmakers, 1973) . A brief description of the elements of this method is given here, and further details are given by Brownlee and Sanger (1969) and Bar ren (1971) . A mini-fingerprinting technique, in which second-dimension homochromatographic separation is performed on small polyethyleneimine (PEI)-cellulose thin-layer plates, has been described by Volckaert et al. (1976) .

Separation in the first dimension is by high-voltage electrophoresis on cellulose acetate strips in 5% acetic acid-0.5% pyridine (pH 3.5), 7 M urea. Oligonucleotides are transferred by blotting from the cellulose acetate strip to a thin-layer plate of DEAE-cellulose. After washing the plate with ethanol to remove urea, second-dimension fractionation is performed by ascending chromatography at 60°C. The Chromatographie solvent, called a homomixture, contains 7 M urea and 5% yeast tRNA (previously partially base hydrolyzed, then dialyzed against water). Other homomixtures, differing in the pretreatment of the tRNA, are used to resolve shorter chain oligonu cleotides. Separation in the second dimension is based upon the relative efficiencies at which the labeled oligonucleotides are displaced by the polynucleotides of the homomixture. Short-chain low-valence oligonucleo tides are more readily displaced along the DEAE stationary phase by the polynucleotides of the mobile phase, and thus migrate more rapidly than oligonucleotides with longer chains and higher valences. As a result, oli gonucleotides are resolved in the second dimension primarily according to their chain lengths. However, the contributions of base composition are important, and pronounced deviations from an inverse relationship be tween mobility and molecular weight are frequently observed.

A rapid one-dimensional fingerprinting procedure has been used to dis tinguish isolates of foot-and-mouth disease virus (La Torre et al., 1982) . RNase TI digests were resolved according to chain length on sequencing gels of 20% polyacrylamide, 8.3 M urea, 1 x 90 mM Tris-90 mM boric acid (pH 8.3)-2 mM EDTA (1 x TBE) (Sanger and Coulson, 1978) . Pat terns were distinguished by the migrations of the larger oligonucleotides as well as by the relative migrations of the tracts of polycytidylic acid [poly(C)], a genomic feature peculiar to the aphthovirus and cardio virus groups. Band patterns of closely related viruses were very similar, but isolates with dis tinct two-dimensional fingerprints also had different one-dimensional band patterns. Although much less sensitive to small genetic differences than twodimensional procedures, this approach may be generally useful as a prelim inary screen for genetic relatedness.

After electrophoresis, one plate is removed and the gel is covered with polyethylene wrap. A screen-type X-ray film, 35 x 43 cm (e.g., DuPont Cronex 4, Kodak XRP, Kodak XAR, or Fuji RX), is placed over the gel. The film may be preflashed to improve sensitivity (Laskey, 1980) . A cal cium tungstate-type intensifying screen (e.g., DuPont Lightning Plus or Fuji Mach 2) is placed over the film, and a Plexiglas plate (36 cm x 43 cm x 1/4 in.) is set on top to assure good contact between the gel and the film. The assembly is securely taped to prevent accidental displacement of the film along the gel, and sealed in a black plastic bag (Picker International) for exposure at 4°C. Good fingerprints are obtained with picornavirus RNA (approximately 7500 bases) after 16 hr of exposure when 5 x 10 6 dpm are applied to the gel. Because diffusion of the larger oligonucleotides in 22% polyacrylamide is very slow at 4°C, gels can be exposed for up to 2 weeks without detectable loss of resolution. Well-defined fingerprints can thus be obtained with as little as 10 5 dpm of RNA.

If the oligonucleotides are to be analyzed further, reference spots of 32 Plabeled radioactive ink are made on the polyethylene sheet at each corner of the gel and covered with clear tape. Oligonucleotides, located by align-ment of reference spots on the developed film and the gel, are excised with a sterile scalpel or cork borer and transferred to a 1.5-ml Eppendorf tube. Oligonucleotides can be recovered with good yields ( > 80%) by mechanical extraction. The brittle second-dimension gel piece is crushed with a flamesealed Eppendorf pipet tip (bulb diameter 2-3 mm) attached to a glass rod (Lee and Wimmer, 1976) . Excessive grinding should be avoided to prevent substantial release of linear pplyacrylamide, which copurifies with nucleic acids. The crushed gel is soaked for 2 hr at 37° C in 750 /xl of elution buffer [0.5 M NH 4 OAc, 10 mM Mg(OAc) 2 , 0.1% SDS, 2 mM EDTA] (Maxam and Gilbert, 1980) containing 50 /xg of carrier tRNA. The gel particles are removed by centrifugation (500 g, 2 min) through a spin filter (see below), washed with 200 μ\ of elution buffer, and centrifuged again. To the filtrate is added 2.5 ml of cold 95% ethanol. After 20 min in a dry ice-ethanol bath, the precipitate is pelleted in a swinging-bucket rotor (10,000 g, 4°C, 30 min), washed with 95% ethanol, centrifuged again (10,000 g, 4°C, 5 min), and lyophilized.

The spin filter set consists of a 1-ml Eppendorf pipet tip plugged with silanized glass wool [Alltech Associates, or prepared as described by Maxam and Gilbert (1980) ], mounted through an adapter to the top of a 12 x 75 mm polypropylene tube. The reusable adapters are made from 1.5-ml Ep pendorf tubes from which the bottoms and the snap caps have been re moved.

A different approach is described by Stewart and Crouch (1981) . The entire fingerprint is transferred to DEAE paper by uniform suction of elu tion buffer through the gel. After drying, the oligonucleotides are stably associated with the DEAE paper, and the transferred fingerprints are vis ualized by autoradiography with intensifying screens at -70°C. Individual oligonucleotides are excised from the DEAE paper and efficiently recovered by elution with triethylamine carbonate, which is removed by lyophilization.

The base compositions of uniformly labeled oligonucleotides can be de termined after complete secondary digestion with RNase A. The double digestion products have the general structure (Ap) 0 -"Xp, where X = C, G, or U. Because Gp is present only at the 3' terminus of an RNase Tl-gen erated oligonucleotide, only one G-containing product should be observed upon secondary digestion, with all other base residues present in integral molar ratios relative to Gp. Two basic approaches are most commonly used to identify the products of secondary digestion. The first, described in detail by Barrell (1971) , utilizes high-voltage electrophoresis on DEAE paper at pH 3.5. The products are identified by their mobilities relative to markers of known composition. Identification of longer products (three or more A residues) can be confirmed by elution from the DEAE, digestion with RNase T2 (non-base specific), and electrophoresis as before at pH 3.5. In the sec ond method, developed by Volckaert and Fiers (1977) , two-dimensional Chromatographie separation occurs on small (6.7 x 10 cm) PEI-cellulose thin-layer plates (CEL 300, Machery-Nagel, Brinkmann). Samples are spot ted 1 cm from each edge and resolved in two steps for each dimension. Elution in the first dimension (along the short axis) is first with water until the front moves 1.5 cm; then the plate is resolved with 20-22% formic acid until the solvent reaches the top. After the plate is air dried, second-di mension separation is at pH 4.3, first in 0.1 M formic acid-pyridine (to 2 cm), then in 1.0 M formic acid-pyridine until the front reaches the top. In the acidic conditions of the first dimension (pH<2), Ap and Cp have no net negative charges and thus have minimal ionic interaction with the PEI support. Because migration rates increase with decreasing negative charge, the digestion products are resolved primarily according to the identities of their 3'-terminal bases, in the order C p > G p > U p . At pH 4.3, migration of the products is reduced with increasing chain length. Thus the base com position of a digestion product can be deduced from its position on the chromatogram.

For determination of sequences, either 5 ' -or 3'-32 P end-labeled oligonucleotides can be utilized in the technique of partial digestion with basespecific RNases (Simoncsits et al., 1977; Donis-Keller et al., 1977; Donis-Keller, 1980) . A commercial kit for enzymatic sequencing of RNA is available from P/L Biochemicals.

Sequences can also be determined from oligonucleotides labeled at either the 5'-or 3'-ends by fingerprinting the products of partial alkaline hy drolysis (wandering-spot analysis). The techniques of separation in wan dering-spot analysis are essentially the same as described for total genome fingerprinting and can utilize either electrophoresis-homochromatography (Rensing and Schoenmakers, 1973) or two-dimensional polyacrylamide gel electrophoresis (Nomoto et al., 1981) . By examining partial hydrolyzates, wandering-spot analysis can reconstruct the effects on the mobility of an oligonucleotide upon stepwise removal of each base residue. Because, for first-dimension mobilities, removal of a cytosine residue causes the greatest shift toward the anode and removal of a uridine residue causes the greatest shift toward the cathode, wandering-spot analysis is most efficient at lo cating pyrimidines. Enzymatic sequencing, in contrast, most readily distin guishes purine residues, because of the base specificities of the RNases Tl (G) and U2 (A + G). The complementary strengths of the two sequencing methods may be combined to sequence oligonucleotides (Darlix et al., 1979; Rommelaere et al., 1979; Nomoto et al., 1981) . Adenosine residues are lo cated by partial cleavage with RNase U2 and separation on sequencing gels. The same partial hydrolyzate used to form the "ladder" in the sequencing gel is also fingerprinted, and the positions of C and U residues are deter mined.

The third direct RNA sequencing technique, involving base-specific chemical cleavages, may also be used, but determinations can be performed only with oligonucleotides labeled at their 3'-ends (Peattie, 1979 

Resolution of individual genome segments, subgenomic transcripts, or full-length viral genomes from defective-interfering RNA species for sep arate fingerprint analysis is usually achieved by gel electrophoresis. The choice of gel system depends upon the size range of the RNA molecules and whether the chains are single or double stranded.

Large, single-stranded RNA molecules over the size range of 0.5-4 x 10 6 can be separated on preparative gels of the high-resolution acid agaroseurea system (Lehrach et al., 1977) . The gel contains 1.5% low-gelling-temperature agarose (Marine Colloids, distributed by Miles Laboratories), 6 M urea, and 25 mM sodium citrate (pH 3.5). The electrode buffer is 25 mM sodium citrate (pH 3.5). If the gels are run horizontally, 6 M urea is in cluded in the electrode buffer. Better resolution is generally obtained with vertical gels, cast between frosted glass plates and supported by a 5% polyacrylamide plug. Electrophoresis is at 4°C at a potential of 2-5 V/cm. RNA is eluted from gel slices by dilution in 5-10 volumes of 20 mM Tris-HC1 (pH 7.8), 1 mM EDTA, and melting the mixture at 65°C for 5 min. RNA is recovered from the mixture by phenol extraction, reextraction of the interphase, and ethanol precipitation of the combined aqueous phases (Weislander, 1979) . If the RNA is to be subsequently labeled in vitro, oyster glycogen (Calbiochem) can be substituted for tRNA as a carrier. Two micrograms of glycogen is used for each microgram of tRNA.

Single-stranded RNA molecules over the size range 0.3-1.5 x 10 6 (e.g., influenza virus genome segments) can be resolved on slab gels containing 2-3% polyacrylamide (acrylamide:Bis 20:1), 6 M urea, with both the gel and electrode buffers consisting of Loening's E buffer [36 mMTris-30 mM NaH 2 P0 4 (pH 7.8)-l mM EDTA] (Scheurch et al., 1975; Loening, 1969) . RNA can be recovered from the gels by mechanical extraction essentially as described for isolation of oligonucleotides.

Good resolution of double-stranded RNA molecules is obtained on a preparative scale by electrophoresis on gels containing 7.5% acrylamide, 0.2% Bis, 6.3 Murea, 0.1 % SDS, in Loening's E buffer. The electrode buffer is Loening's E buffer containing 0.1% SDS (Ito and Joklik, 1972) . The discontinuous SDS-polyacrylamide gel system of Laemmli (1970) may also be used for preparative isolation of double-stranded RNA molecules (Clewley and Bishop, 1979a) . Double-stranded RNA molecules may be recovered from polyacrylamide gels by mechanical extraction or by electrocution into dialysis membranes (Scheurch et al., 1975; McDonell et al., 1977) . Gels are soaked for several hours in deionized water to remove SDS and stained for 30 min with 10 jug/ml ethidium bromide in 0.2x TBE. The RNA bands, visualized under long-wavelength ultraviolet light, are excised from the gel. Each gel slice is placed in a dialysis bag (previously boiled in 5 mM EDTA, then autoclaved for 10 min in deionized water) containing a small volume of 0.2 x TBE, and the bag is sealed at both ends without trapping air bub bles. Bags are placed longitudinally between the electrodes of an electroelution tank containing just enough 0.2 x TBE to submerge the bags. Elec trophoresis is at 100 V (0-4°C) until the ethidium bromide-stained RNA migrates from the gel (usually 2-3 hr) and forms a fluorescent streak along the inside of the bag. The polarity of electrophoresis is reversed for 2 min, the buffer is removed from the bag, and the bag is washed once with 0.2 x TBE. The RNA-containing buffer is twice extracted with phenol, and the RNA is recovered by ethanol precipitation.

Ortin et al. (1980) isolated influenza virus segments for fingerprinting by hybridization of total viral RNA to plasmid DNA, immobilized on nitro cellulose filters, containing the sequences of specific segments. After hy bridization at 37°C for 24-36 hr in 50% formamide, 0.5% SDS, 5 x SSC (1 x SSC = 150 mM NaCl, 15 mM sodium citrate, pH 7.0), the filters were washed with 2 x SSC, and the hybrids were treated with 10 units/ml RNase Tl in 10 mMTris-HCl (pH 7.5), 0.5 M NaCl, in the presence of 20 μεΐπΰ carrier tRNA. The filters were again washed with 2 x SSC, and the resistant RNA was eluted at 50° C in formamide and recovered by ethanol precipi tation. Hybridization selection should find wider application as a prep arative method with the increasing availability of cloned copies of viral genes.

G. OLIGONUCLEOTIDE MAPPING

Physical genomic maps of RNA viruses can be constructed by determi nation of the order of the unique RNase Tl oligonucleotides along the RNA molecule. Ordering of oligonucleotides is most easily accomplished with 2.0 . For these, the basic analytical strategy is partial fragmentation of the viral RNA, selection of the poly(A)containing fraction, size fractionation of the coterminating fragments, and fingerprinting to determine which oligonucleotides disappear with decreasing chain length (Wang et al., 1975) . Several conditions have been used to produce partial fragmentation of the viral RNA, including limited base hydrolysis (Wang et al., 1975) , sonication (Kennedy, 1976) , boiling in Tris-EDTA buffer (Merregaert et al., 1981) , partial digestion with RNase III (Stewart et al., 1980) , or radioactive decay of 32 P-labeled phosphodiester bonds (Faller and Hopkins, 1978) . Nested sets of polynucleotides are generally obtained with common 3 ' -ends by se lection of the poly(A)-containing fraction on oligo(dT)-cellulose (Aviv and Leder, 1972) or poly(U)-Sepharose (Lindberg and Persson, 1972) . Harris et al. (1980) obtained fragments of varying lengths extending from the poly(C) tract near the 5 ' -end of foot-and-mouth disease virus RNA by selection on oligo(dG)-cellulose. Separation according to chain length is generally per formed in denaturing sucrose gradients (Wang et al., 1975; Darlix et al., 1979) . Better separation of individual size classes of fragments is achieved by electrophoresis on SDS-polyacrylamide-agarose gels (Stewart et al., 1980) , but sedimentation gradients are usually preferred because the frac tions are more easily recovered.

Results of a typical mapping experiment, following the procedures of Wang et al. (1975) , are shown in Fig. 5 . The basic steps are given below. Partial base hydrolysis of in vivo 32 P-labeled virion RNA, incubated at 50°C in low-salt buffer [LSB; 10 mM Tris-HCl (pH 7.8), 10 mM NaCl, 1 mM EDTA], is brought about by addition of 0.05 volume of 1.0 M Na 2 C0 3 to increase the pH to 10.8. At various intervals (e.g., 0, 15 sec, 30 sec, 1 min, 2 min, . . .) small aliquots are neutralized in 0.1 volume of 0.5 M acetic FIG. 5 . Mapping of the oligonucleotides of the Leon strain of type 3 poliovirus using the methods of Wang et al. (1975) and Nomoto et al. (1979b) . Virion RNA labeled in vivo was partially base hydrolyzed, and the poly(A)-containing RNA fragments were selected by two cycles of affinity chromatography on oligo(dT)-cellulose. The poly(A)-containing fragments were separated according to size by sucrose gradient sedimentation, followed by resedimentation of each of the 10 pooled fractions. Peak fractions from each gradient were fingerprinted (B-J). The fraction A fingerprint is not shown because it was identical to that of untreated virion RNA (Fig. 1) . Spots are numbered as in Fig. 1. (Lower panel) A map of oligonucleotide order determined by visual comparisons of relative spot intensities in the fingerprints above. Numbers given in brackets are arranged in sets of ascending values according to the intensities of the spots in the next (lower S value) gradient pool. Spots that are missing in the next fin gerprints are listed first; those that are present but visibly lighter than the others are given in succeeding sets depending upon intensity. The overall map is compressed relative to the phys ical genome, probably because the relationship between genome size and experimentally de termined S value deviates from the equation of Spirin (1963) used to calibrate the map. acid. Optimal incubation and sampling conditions vary with different RNA preparations. The neutralized samples are then pooled, and ethanol pre cipitated. To select for the poly(A)-containing fraction, the RNA is resuspended in 500 μΐ of oligo(dT) binding buffer [10 mM Tris-HCl (pH 7.8), 0.5 MNaCl, ImMEDTA, 0.5% SDS], applied to a column containing 500 mg of oligo(dT)-cellulose, and the column is washed with 20 volumes of binding buffer. Poly(A)-containing RNA is eluted from the oligo(dT) by addition of elution buffer [10 mM Tris-HCl (pH 7.8), 1 mMEDTA, 0.05% SDS], and the peak fractions are pooled. Two cycles of oligo(dT) selection are generally required to eliminate non-poly(A)-containing RNA contami nants. After ethanol precipitation, the poly(A)-containing RNA fragments are resuspended in 0.5 ml of LSB/0.1% SDS, heated at 70°C for 2 min, and resolved by sedimentation at 20°C in a linear 15-30% sucrose gradient in LSB/0.1% SDS. Pools of gradient fractions may be fingerprinted di rectly. However, resedimentation of each gradient fraction pool may be necessary to obtain greater size homogeneity within each sedimentation fraction (Nomoto et al., 1979b) . In some procedures, oligo(dT) selection is the final step before fingerprinting, with each size class pool selected in small Pasteur pipet columns containing approximately 100 mg of oligo(dT).

Oligonucleotide maps may be constructed by visual comparison of the fingerprints of each size class and recording the order in which oligonu cleotide spots become weak and then absent with decreasing fragment size. More quantitative comparisons are made by fingerprinting digests of each fragment size class mixed with differentially labeled full-length RNA. The RNA fragments are labeled with [ 3 H]uridine and the full-length reference RNA with 32 P. The unique oligonucleotides are excised from the finger prints, and their radioactivity is determined by scintillation counting. The order of the oligonucleotides is deduced by plotting the 3 H/ 32 P ratios for each oligonucleotide as a function of the size fractionation range of the gradients (Coffin et al., 1978) .

Because of the limited resolving power of sedimentation gradients and the nonuniform distribution of unique oligonucleotides along the genome, the maps obtained by the above procedures are only approximate and are not a precise measure of the physical distances between oligonucleotides. The largest fragments are least efficiently resolved because they have the most similar sedimentation values; therefore the largest errors in assigning the oligonucleotide order are expected to occur for the 5'-terminal region. Oligonucleotide maps have been useful in identifying and characterizing recombinant genomes produced by intramolecular exchange (Beemon et al., 1974) , in correlating biological properties with the presence of certain mapped oligonucleotides (Wang, 1978) , in mapping deletion mutants of viral genomes (Kennedy, 1976) , and in localizing some mutational differ ences between closely related strains (Nomoto et al., 1981) .

Oligonucleotides of the negative-stranded viruses vesicular stomatitis vi rus (Freeman et ai, 1979; Clewley and Bishop, 1979a) and Sendai virus have been assigned to specific structural genes by identifying the set of oligonucleotides protected from nuclease digestion in mRNA-virion RNA duplexes. Freeman et al. (1979) separately hybridized 32 P-labeled virion RNA to saturating amounts of each viral mRNA. The partial duplexes were digested with RNase Tl and finger printed, and the oligonucleotides appearing in the fingerprints were as signed to the gene encoding the missing mRNA. Clewley and Bishop (1979a) applied the reciprocal approach and produced separate duplexes with each mRNA, then digested the single-stranded regions with a nuclease mixture containing SI nuclease and RNases A, T l , and T2. The protected duplexes were purified on polyacrylamide gels and fingerprinted. The observed oli gonucleotides were assigned to the gene encoding the protecting mRNA. Because the structural genes had been ordered by independent means, a partial oligonucleotide map could be constructed. Further ordering of oli gonucleotides within each gene would require partial hydrolysis and fractionation of the poly(A)-containing mRNAs. This approach was useful in mapping deletions in defective-interfering mutants (Freeman et al., 1979; and in comparing the genes of closely related rhabdoviruses (Clewley and Bishop, 1979a) .

Fingerprinting has been used to provide a measure of the number of nucleotide residues per unique genome. The analysis is based on the principle that, for uniformly labeled RNA, the specific activity of any unique oli gonucleotide is equivalent to that of the complete RNA molecule (Sinha et al., 1965; Fiers et al., 1965) . Thus the chain length per unique genome can be calculated from the relationship , . , , , chain length (oligonucleotide) x cpm (RNA) chain length (RNA) = cpm (oligonucleotide)

In practice, uniformly labeled viral RNA is fingerprinted and several unique oligonucleotides are analyzed for radioactivity and chain length, de termined from either composition data (Billeter et al., 1974) or sequence data (Lee et al., 1979) . Several oligonucleotides are examined because of variable recoveries of oligonucleotides from the fingerprints. Some losses of the input RNA counts occur during fingerprinting, and correction for losses has been approached by the addition of oligonucleotides of known specific activities (internal standards), by parallel analyses of RNAs of known size and complexities (external standards) (Billeter et al., 1974) , or by counting the entire fingerprint gel (Lee et ai, 1979 ).

The complexity, or level of ploidy, of an RNA viral genome can be es tablished by comparing the minimum genome size with other physical mea sures of the size of the viral RNA. Such analyses have been widely used to determine the complexities of the polyploid retroviral genomes (Billeter et ai, 1974; Beemon et ai, 1976; Friedrich et al., 1976; Vigne et al., 1977) . Complexity analysis has also been used to determine the chain lengths of the genomic RNAs of Semliki Forest virus (Lomniczi and Kennedy, 1977) , avian infectious bronchitis virus (Lomniczi and Kennedy, 1977) , poliovirus types 1 and 2 (Lee et al., 1979) , and the genome segments of the bunyaviruses Uukuniemi virus (Petterson et al., 1977) and snowshoe hare virus (Clewley et al., 1977b) .

The ultimate identification of a virus strain is the determination of its complete genomic sequence. In studies to determine the total genome se quence of the Mahoney strain of type 1 poliovirus , RNase TI-and RNase A-resistant oligonucleotides were used to prime the synthesis of DNA chains in the DNA polymerase I-mediated dideoxy se quencing system (Kitamura and Wimmer, 1980) . Because the viral genome and the oligonucleotide primers are of the same polarity, the template was cDNA synthesized from virion RNA. The resultant sequence, unlike those determined from cloned DNA copies, represents the average genome se quence of the viral population (see the next section). Oligonucleotides have also been used in sequence studies with cloned DNA, serving as hybridi zation probes to characterize clones (Boothroyd et al., 1981; van der Werf étal., 1981) .

The fingerprint obtained experimentally represents the average genotype of a large virus population. Because of the extremely high inherent muta bilities of RNA genomes (Holland et al., 1982) , substantial genetic poly morphism may exist in the virus population. Model fingerprinting studies with phage Q/3 revealed that the genome of most individual members of the population differed from the average genome at one or more sites (Dom-ingo et al., 1978) . Further, more than one average genome may exist for the population, particularly if the virus population is in a state of disequi librium (Holland et al., 1982) . If the alternative average genomes have dif ferent fingerprints, then the variant oligonucleotides will be detected as lighter, submolar spots. It is the experience of many laboratories that virus stocks maintained in the laboratory by low-multiplicity passage in cell cul ture appear to maintain a stable genetic equilibrium, such that their fin gerprints remain unaltered with increased passage.

During natural transmission a virus population may encounter numerous complex conditions that favor disequilibrium and the selection of new var iants. Given a succession of selective conditions, genome evolution is con tinual and is detected as cumulative changes in the fingerprints of successive isolates. The nature of the selective pressures is poorly defined, but appears to involve factors in addition to host antibodies, as natural genomic evo lution can involve changes in sequences encoding nonstructural proteins (Young et al., 1979; Domingo et al., 1980; Nottay et ai, 1981) and can occur during persistent infection of cultured cells (Holland et ai, 1979; Meinkoth and Kennedy, 1980) and during infection of persons with severe immunodeficiencies (Yoneyama et al., 1982) .

In fact, the observed rates of genome evolution in nature vary widely among RNA viruses. Presumably, this is due to varying degrees of selective pressure encountered during the natural life cycles, as well as to possible differences in the basic mutation rate (Pringle et al., 1981) .

For some viruses, e.g., influenza A virus (Young et al., 1979; , foot-and-mouth disease virus , and poliovirus , the rate of natural evolution is so rapid that it is possible to reconstruct the general pathways of epidemiological trans mission based upon the pattern of fixation of changes into the fingerprints. Such applications of fingerprint analysis have sometimes been described as "molecular epidemiology.'' Many other viruses, including vesicular stomatitis virus (Clewley et al., 1977a) and Western equine encephalitis virus (Trent and Grant, 1980) , evolve so slowly that viruses with similar fingerprints can be isolated over a period of many years. In such cases, transmission pathways cannot be determined by fingerprint analysis, although the distribution of a genotype over space and time may be followed. Many virus strains appear to be re stricted geographically, possibly because of the limited range of their res ervoirs and vectors, and disappear from the environment as a genotype identifiable by fingerprinting, either by continued evolution or by displace-ment by strains of the same group but with distinct fingerprints. Virus gen otypes, recognized by fingerprinting and geographic distribution, are called "topotypes."

Representatives of nearly every RNA virus group have been finger printed. The available data constitute a growing fingerprint catalog of RNA virus genotypes. For some viruses, the catalog may require frequent up dating, because of the rapid evolution, whereas for others updating may not be necessary for decades.

1. Molecular Epidemiology a. Influenza Viruses. Fingerprint analysis of influenza A (HlNl) iso lates from the 1977-1979 pandemic revealed their close genetic relationship to H l N l strains isolated in 1950 (Nakajima et al, 1978) . The 1977 The -1979 isolates showed a pattern of continual evolution by fixation of mutations, allowing construction of an evolutionary tree based upon fingerprint changes (Young et al., 1979) . In view of the rapid evolution of the pandemic virus, the great similarities in the fingerprints of isolates obtained 27 years apart suggested that the 1950 virus had reemerged after being maintained in a state of genetic dormancy, the exact nature of which is unknown. The evolution of H3N2 viruses has also been studied by fingerprinting (Ortin et ai, 1980; Nakajima et al., 1982) .

Influenza C viruses appear to be much more stable in nature, as isolates from four continents obtained over a 32-year period had generally similar fingerprints (Meier-Ewert et al., 1981) .

b. Enteroviruses. We use fingerprinting in our laboratory as a routine method for identifying the poliovirus genotype associated with an epidemic. The results of an epidemiological study are shown in Fig. 6 . It has been possible, on the basis of fingerprint data alone, to reveal unsuspected epidemiological links between cases or outbreaks that occur at different locations and at different times . An impor tant fraction of isolates from the few remaining cases of paralytic polio myelitis in developed countries has been shown, by fingerprinting, to be genetically related to the live poliovaccines (see, e.g., Fig. 2) (Minor, 1980; Nottay et al., 1981; Kew et al., 1981) . Case isolates of enterovirus 70 from the 1980-1981 pandemic of acute hemorrhagic conjunctivitis were shown by fingerprinting to be closely related to each other and to some prepandemic isolates (Kew et al., 1983) .

c. Foot-and-Mouth Disease Viruses (FMDV). Isolates of type C FMDV obtained from diseased swine and cattle in Spain in 1979 were shown to be genetic variants of a 1970 strain from the same region . Similar findings were presented by this group for a type O strain. FMDV isolates from a 1981 outbreak in England and France were shown to have fingerprints nearly identical to that of a vaccine strain isolated in 1965. Because of the high mutability of FMDV during epizootic transmission, unaltered survival of the 1965 strain in the field is unlikely, strongly suggesting that the outbreak was caused by accidental reintroduction of the virus into the field (King et al., 1981) .

at Different Times and Locations a. Picornaviruses. Frisby et al. (1976) fingerprinted representatives of three of the four major picornavirus subgroups. Except for the cardioviruses encephalomyocarditis virus and Mengo virus, which had similar fin gerprints, different viruses within and across subgroups had dissimilar fingerprints. The genetic relationships between isolates of swine vesicular disease virus and coxsackievirus B5 were studied by fingerprinting and com petition hybridization . Fingerprint analysis showed that the eight independent isolates of Theiler's murine encephalomyelitis virus belong to two distinct genetic subgroups (Lorch et al., 1981) . Attenuated strains of FMDV and poliovirus (Nomoto et al., 1981; Nottay et al., 1981) have been compared by fingerprinting with their virulent parents.

b. Alphaviruses. The alphaviruses have undergone extensive evolution ary divergence, as little sequence homology is detected by hybridization across the three serocomplexes, Eastern equine encephalitis (EEE), Western equine encephalitis (WEE), and Venezuelan equine encephalitis (VEE) (Wengler et al., 1977) . For both the VEE (Wengler et al., 1977; and WEE (Trent and Grant, 1980) groups, fingerprint analysis generally corroborated serological classification. Isolates that were very close serologically had similar fingerprints, clearly distinguishable from those of close, but distinct, serological relatives. Individual alphavirus strains appear to evolve slowly, as very similar fingerprints were charac teristic of WEE isolates widely separated temporally and geographically (Trent and Grant, 1980) . c. Flaviviruses. The flavi viruses are divided into seven subgroups, with very limited genetic homology across subgroups. Fingerprints of viruses of different subgroups (Wengler et al., 1978) or different viruses within a subgroup (the prototypes of dengue virus types 1, 2, 3, and 4) (Vezza et al., 1980a) showed no similarities. However, when comparisons were re stricted to independent isolates of the same flavi virus, substantial genetic homology among representatives of a strain was evident. Trent et al. (1981) surveyed numerous North and Central American isolates of St. Louis en cephalitis (SLE) virus from humans, birds, rodents, and mosquitoes ob tained over a 47-year period. Fingerprint similarities were seen among all isolates, with differences among genotypic variants primarily correlating with differences in geographic region of isolation.

d. Rhabdoviruses. Fingerprints of the genetically divergent rhabdoviruses vesicular stomatitis virus (VSV), rabies, Cocal, Chandipura, and spring viremia of carp virus were, as expected, quite distinct (Clewley and Bishop, 1979b) . However, independent isolates of the VSV Indiana group, obtained from diverse biological sources at different locations over a 50-year period, shared many similarities in their fingerprints (Clewley et al, 1977a) . The existence of two genetic subtypes of VSV New Jersey was demonstrated by fingerprint analysis. Three different laboratory strains (Clewley and Bishop, 1979b) of rabies have been fingerprinted.

e. Bunyaviruses. The bunyavirus group contains over 200 accepted or proposed members, divided among four genera. Members of different serogroups have clearly unique fingerprints for each L, M, and S genome seg ment (Pettersson et al, 1977; Ushijima et al, 1980; Clerx and Bishop, 1981) . Within the California serogroup, individual strains, distinguishable on serological grounds, were also shown to have dissimilar fingerprints (El Said et al, 1979) . Fingerprint similarities were evident for independent isolates of La Crosse virus (California serogroup), although differences were ob served among isolates. Variants existed as topotypes, as isolates obtained over 16 years from the same region were more closely related to each other than isolates obtained at the same times from different regions. The genome segments of La Crosse appeared to evolve at different rates, M > L > S (El Said et al, 1979) .

/ . Arenaviruses. Two variants of Pichinde virus (Tacaribe serocomplex) isolated 5 years apart in Colombia had clearly distinguishable fingerprints for their respective L and S segments. Vezza et al (1980a) utilized these differences to provide physical evidence for independent assortment among arenaviruses.

g. Coro navi ruses. The fingerprints of seven isolates of mouse hepatitis virus, obtained at different times and locations, indicated that all members of the group were closely related, having from 40 to 95% of their oligonucleotides in common . Independent isolates of infectious bronchitis viruses (IBV) had many distinct fingerprints, indicat ing a potential for substantial genetic variation even within a serotype. At the same time, some genotypes of IBV appeared to have wide geographic and temporal distributions (Clewley et al, 1981) .

h. Measles Virus. Fingerprints of measles virus isolated from cases of acute disease and from cases of subacute sclerosing panencephalitis revealed distinguishable patterns for each, with some large oligonucleotides shared (Stephenson and ter Meulen, 1982) .

/'. Ebola Virus. Coetaneous isolates of Ebola virus, from patients with hemorrhagic fevers, obtained in different locations (Zaire and Sudan) had very distinct fingerprints. Subsequent isolates had fingerprints similar to the earlier strains found in each location (Cox et al, 1983) , suggesting a slow rate of evolution for each Ebola topotype. j . Orbiviruses. Fingerprint comparisons of isolates of bluetongue virus type 11 (BTV-11), obtained over a 12-year period in the United States, in dicated that evolution in the natural environment occurred by both mutational drift and reassortment, and that several BTV-11 genotypes could be in circulation in the United States at the same time (Sugiyama et al., 1982) . Two Australian orbiviruses have also been compared by fingerprinting (Walker et al., 1980) . k. Retroviruses. Fingerprinting has found wide application in the study of retroviruses. The large literature on this subject was separately and com prehensively reviewed by Beemon (1978) and Wang (1978) . More recently, fingerprinting was used to reveal the genetic similarity of viruses from hu man leukemia patients to simian and gibbon ape retroviruses (Sahagan and Haseltine, 1980) , to demonstrate that retroviral sequences from lymphomas of AKR mice are produced by recombination among at least three different endogenous retroviruses (Pedersen et al., 1981; Thomas and Coffin, 1982) , to document the participation of retroviral mRNA molecules in the recom bination process (Wang and Stacey, 1982) , and to show that retroviruses isolated from wild mice in different geographic locations had divergent se quences (Lai et ai, 1982) . Using a combination of fingerprinting and oligonucleotide sequencing, Clements et al. (1982) demonstrated that common nucleotide sequence changes occurred in antigenic variants of visna virus isolated from different sheep.

C. OTHER APPLICATIONS RELATING TO VIRUS IDENTIFICATION

Recombination by intramolecular exchange was demonstrated for retro viruses (Beemon et al., 1974) and picornaviruses (King et al., 1982) by fingerprint analysis of those progeny from mixed infections having recombinant genetic markers. In both systems, parental strains having eas ily distinguishable fingerprints and well-defined oligonucleotide maps were used. The approximate sites of crossover could be mapped because, for any region along the genome, only one parent contributed oligonucleotides to the recombinant RNA molecule. Natural intertypic recombinants of the live poliovaccines, isolated from humans exposed to trivalent oral vaccine, have been characterized by oligonucleotide mapping (Kew and Nottay, 1984) .

Genetic exchange in viruses with segmented genomes occurs by independent assortment. Although reassortment can be confirmed by other physical means, such as one-dimensional electrophoresis (Young and Palese, 1979; Walker et al., 1980) , fingerprinting of individual segments is much more sensitive and can clearly indicate whether each segment is unique. Finger printing has been used to characterize natural reassortants of influenza A viruses (Young and Palese, 1979; Nakajima et al., 1982) , bluetongue virus (Sugiyama et al., 1982) , and the bunyavirus Patois (Ushijima et al., 1980) .

Defective-interfering (DI) particles containing genome deletions are pro duced by nearly all RNA viruses. The presence of DI RNAs may complicate comparative studies among viruses because they can contribute substantial heterogeneity to the virus RNA population. Deletions may be located by comparison of the fingerprints of DI RNAs with the mapped oligonucleotides of full-length genome. Defective RNA genomes of avian retroviruses (Wang et al., 1975; Beemon, 1978) , Semliki Forest virus (Kennedy, 1976; Pettersson, 1981; Soderlundeitf/., 1981) , Sindbis virus (Dohnerei al., 1979) , poliovirus (Nomoto et al., 1979b) , influenza virus (Davis and Nayak, 1979; Nakajima et al., 1979) , vesicular stomatitis virus (Clerx-van Haaster et al., 1980; Hagen and Huang, 1981) , and Sendai virus have been characterized by fingerprinting.

Subgenomic RNA molecules of genome polarity are produced intracellularly during the infectious cycles of alphaviruses, retroviruses, and coronaviruses. Fingerprinting has been used to map the subgenomic RNAs to the full-length genome for Semliki Forest virus (Kennedy, 1976; Wengler and Wengler, 1976) , Rous sarcoma virus (Mellon and Duesberg, 1977) , infec tious bronchitis virus (Stern and Kennedy, 1980) , and mouse hepatitis virus .

Fingerprinting is a definitive method of identifying RNA viruses accord ing to their genotypes. It is not subject to the problems of antigenic drift or antigenic convergence that complicate serological identification. Fur thermore, it provides a semiquantitative means of following the evolution of viral genomes in nature. Because all regions of the genome are repre sented by the large diagnostic oligonucleotides, a survey of the total genomic changes can be monitored.

Fingerprinting has two limitations as a diagnostic tool. First, although highly definitive, fingerprinting is not as rapid or inexpensive as serological techniques and cannot be as easily scaled up for routine identification of a large number of samples. Second, the evolutionary range of fingerprinting is short, and relationships may not be evident for isolates of rapidly evolv ing viruses obtained over long intervals. However, these limitations are not large, compared to the full benefits offered the virologist by the finger printing method.

Procedures in Nucleic Acids Research

New Developments in Practical Virology

Modern Approaches to Vaccines

New Developments in Practical Virology

Virology

' T h e Enzymes

We are grateful to Drs. Fred Brown, Tim Harris, Yuan Fon Lee, and Eckard Wimmer for introducing our laboratory to the fingerprinting technique. We thank Marie Knox for careful preparation of the manuscript.