Journal of General Virology

Molecular intermediates of fitness gain of an RNA virus: characterization of a mutant spectrum by biological and molecular cloning

Armando Arias,¹ Ester Lázaro,² Cristina Escarmís¹ and Esteban Domingo¹

¹ Centro de Biología Molecular 'Severo Ochoa' (CSIC-UAM), Universidad Autónoma de Madrid, Cantoblanco, 28049 Madrid, Spain
² Centro de Astrobiología (CSIC-INTA), Carretera de Ajalvir, km 4, 28850 Torrejón de Ardoz, Madrid, Spain

The mutant spectrum of a virus quasispecies in the process of fitness gain of a debilitated foot-and-mouth disease virus (FMDV) clone has been analysed. The mutant spectrum was characterized by nucleotide sequencing of three virus genomic regions (internal ribosome entry site; region between the two AUG initiation codons; VP1-coding region) from 70 biological clones (virus from individual plaques formed on BHK-21 cell monolayers) and 70 molecular clones (RT–PCR products cloned in E. coli). The biological and molecular clones provided statistically indistinguishable definitions of the mutant spectrum with regard to the distribution of mutations among the three genomic regions analysed and with regard to the types of mutations, mutational hot-spots and mutation frequencies. Therefore, the molecular cloning procedure employed provides a simple protocol for the characterization of mutant spectra of viruses that do not grow in cell culture. The number of mutations found repeated among the clones analysed was higher than expected from the mean mutation frequencies. Some components of the mutant spectrum reflected genomes that were dominant in the prior evolutionary history of the virus (previous passages), confirming the presence of memory genomes in virus quasispecies. Other components of the mutant spectrum were genomes that became dominant at a later stage of evolution, suggesting a predictive value of mutant spectrum analysis with regard to the outcome of virus evolution. The results underline the observation that greater insight into evolutionary processes of viruses may be gained from detailed clonal analyses of the mutant swarms at the sequence level.

RNA viruses replicate as complex mutant distributions termed virus quasispecies (Eigen, 1971; Domingo et al., 1978, 2001; Eigen & Biebricher, 1988; Holland et al., 1992; Nowak, 1992). RNA genome replication must be viewed as a dynamic process in which mutants arise at high rates and participate in a continuous process of competitive rating (Batschelet et al., 1976; Domingo et al., 1978, 1988, 2001; Eigen & Biebricher, 1988; Drake & Holland, 1999). One of the consequences of competition among mutant genomes is fitness gain when large populations of RNA viruses are allowed to replicate in a defined environment (Novella et al., 1995 a, b, c; Escarmís et al., 1999; Weaver et al., 1999). Virus fitness is defined as the relative capacity of a virus to produce infectious progeny under a set of environmental conditions. It is necessarily a dimensionless, relative value, which is frequently determined in growth-competition experiments involving co-infection of cells or animals with the virus to be tested and a reference virus (Holland et al., 1991; reviewed in Domingo et al., 2001).

In contrast to large population passages, repeated plaque-to-plaque transfers of RNA viruses result in fitness loss (Chao, 1990; Duarte et al., 1992; Escarmís et al., 1996; Yuste et al., 1999). This is due to the accumulation of deleterious mutations as a result of repeated sampling (bottlenecking) of components of the mutant spectrum of the virus quasispecies. It must be stressed that a virus population may either gain or lose fitness depending on both the initial fitness of the population and the size of the genetic bottleneck, as shown by Novella et al. (1995 c) using vesicular stomatitis virus clones and populations of different initial fitness. In a study with the animal picornavirus foot-and-mouth disease virus (FMDV), a highly debilitated (low fitness) clone was derived by 22 successive plaque-to-plaque transfers of an FMDV clone termed C⁹₁; the debilitated clone was termed C⁹₂₂ and its fitness value was 0.1 times that of C⁹₁ (Escarmís et al., 1996). The consensus genomic nucleotide sequence of C⁹₂₂ differed from that of C⁹₁ in seven point mutations and in the acquisition of an internal polyadenylate tract as an extension of four adenylate residues at genomic positions 1119–1122, preceding the second functional AUG initiation codon (Fig. 1). FMDV genome residues have been numbered according to Escarmís et al. (1996). One of the mutations was a deletion of a U residue at position 1056 (U-1056), which rendered the region between the two AUG initiation codons non-functional with regard to protein-coding, since U-1056 led to a termination codon at the position that would encode the eleventh residue of the large form of L protease (termed Lab) (Escarmís et al., 1996). The internal polyadenylate at genomic positions 1119–1122 was the first genetic lesion to revert to the wild-type sequence in the process of fitness gain when C⁹₂₂ was subjected to large population passages, while U-1056 did not revert, and served as a specific genetic marker for the C⁹₂₂ lineage (Escarmís et al., 1999) (Fig. 1).

Fig. 1. Scheme of the FMDV C-S8c1 genome and the location of some genetic alterations within residues 605–1122 and 3208–3834 in C⁹₂₂ p0, C⁹₂₂ p20, C⁹₂₂ p50 and C⁹₂₂ p100. VPg is the protein covalently linked to the 5´ end of FMDV RNA and (A)_n is the polyadenylate tract at the 3´ end of the genome. The C-S8c1 genome is 8115 nucleotides in length, not counting the heterogeneous internal poly(C) and terminal poly(A) tracts (Toja et al., 1999). The positions of the non-structural (L and 2A to 3D) and structural (1A or VP4 to 1D or VP1) proteins are indicated. Below the C-S8c1 genome, the genetic lesions within residues 605–1122 and 3208–3834 that are relevant to the interpretation of the composition of the mutant spectrum of C⁹₂₂ p50 are indicated. U-1056 is the deletion of U-1056 of the C-S8c1 genome; the deletion is present in and serves as a genetic marker for the C⁹₂₂ lineage. C⁹₂₂ p0 refers to the virus population amplified minimally from about 10⁵ p.f.u. (C⁹₂₂) to about 10⁷ p.f.u. (C⁹₂₂ p0) prior to measurement of fitness and nucleotide sequencing (Escarmís et al., 1996). (A)_n in C⁹₂₂ p0 indicates the heterogeneous internal polyadenylate tract which was the first lesion to revert upon large population passages of C⁹₂₂ p0 (Escarmís et al., 1999) (reversions are indicated by asterisks). Within the 1D (VP1)-coding region, transversion C-3653 A, which leads to amino acid replacement T-149 K, was present transiently in C⁹₂₂ p50 and was then replaced by two neighbouring transversions in C⁹₂₂ p100; amino acid replacements are boxed. The entire genomic consensus nucleotide sequences of C-S8c1, C⁹₂₂ p0, C⁹₂₂ p20, C⁹₂₂ p50 and C⁹₂₂ p100 have been determined (Escarmís et al., 1996, 1999; Toja et al., 1999). For general reviews of the FMDV genome and its expression, see Belsham (1993), Rueckert (1996) and Sobrino et al. (2001); for additional information on the origin of C-S8c1, C⁹₂₂ and derived populations, see Escarmís et al. (1996, 1999).

The change of fitness values from FMDV clone C⁹₁ to C⁹₂₂ and then to C⁹₂₂ p100 (C⁹₂₂ subjected to 100 large population passages in BHK-21 cells) is depicted in Fig. 2. The relative fitness of C⁹₂₂ p100 was about 30-fold that of C⁹₂₂ (Escarmís et al., 1999). In the present report, we analyse the composition of the mutant spectrum of population FMDV C⁹₂₂ p50, positioned half-way in the exponential phase of the process of fitness recovery of clone C⁹₂₂ (Fig. 2). The mutant spectrum has been analysed by two different procedures: nucleotide sequences of genomic RNA from biological clones of C⁹₂₂ p50 (referred to as biological cloning) and nucleotide sequences of molecular clones obtained in E. coli following RT–PCR of RNA from C⁹₂₂ p50 (referred to as molecular cloning). The analysis of the mutant spectrum of C⁹₂₂ p50 had three objectives: (i) to determine the complexity of the mutant spectrum of a virus quasispecies during fitness gain in a constant environment; (ii) to define molecular intermediates of fitness gain in relation to the initial (C⁹₂₂ p0) and subsequent (C⁹₂₂ p100) consensus genomic sequences (Fig. 2); and (iii) to compare the composition of the mutant spectrum as deduced from the analysis of biological clones and molecular clones.

Cells, viruses and infections. Host cells used for all infections were derived from a clone of BHK-21 cells obtained by limiting dilution, as described previously (de la Torre et al., 1988). Procedures for infections with FMDV in liquid culture medium and in semi-solid agar medium for titration of FMDV infectivity were described in detail elsewhere (Sobrino et al., 1983; Escarmís et al., 1996). FMDV C-S8c1 is a biological clone derived from the natural isolate C-Sta Pau Sp/70 (Sobrino et al., 1983) and FMDV C⁹₁ is a clone derived from C-S8c1 passaged twice in BHK-21 cells (Escarmís et al., 1996). The origin of C⁹₂₂ p50 is described in Escarmís et al. (1999) and in Fig. 2. Biological clones from C⁹₂₂ p50 were obtained by isolating virus from randomly chosen individual virus plaques, as described previously (Sobrino et al., 1983). RNA extraction, cDNA synthesis, PCR amplification and nucleotide sequencing were carried out as detailed in Escarmís et al. (1999).

Fig. 2. Change in relative fitness of FMDV clone C⁹₁ upon plaque-to-plaque transfer (left) and large population passage (right) in BHK-21 cells. Plaque isolations (progeny from a single genome) are indicated as filled squares and uncloned populations as open circles. Clone C⁹₁ was obtained from population C-S8c1 p2 and subjected to 22 serial plaque transfers, as detailed in Escarmís et al. (1996). Each large population passage of C⁹₂₂ p0 (the initial preparation derived from clone C⁹₂₂ by amplification from about 10⁵ to 10⁷ p.f.u.) involved infection of 4x10⁶ BHK-21 cells with 10⁶–10⁷ p.f.u. of the virus progeny of the previous infection. The number of p.f.u. employed ensured exponential fitness gain over the relative fitness range covered from passage 0 to passage 100 (Novella et al., 1995 c). Relative fitness values are taken from Escarmís et al. (1999) and were determined by growth-competition experiments of the different populations to be tested and a reference virus derived from C-S8c1, as described previously (Holland et al., 1991; Duarte et al., 1992; Escarmís et al., 1996, 1999). The short horizontal segments above populations indicate that the consensus nucleotide sequence of the entire FMDV genome is known and can be found in Escarmís et al. (1996, 1999). The mutant spectrum of the C⁹₂₂ p50 quasispecies was analysed by biological and molecular cloning (filled rectangles). Additional details of the procedures used are given in Methods.

Molecular cloning. RNA extracted from C⁹₂₂ p50 was amplified by RT–PCR using the thermostable Pfu polymerase (Promega), which has a proof-reading activity (Cline et al., 1996). The synthesis of cDNA was carried out with 5 U AMV RT (Promega) in a final volume of 25 µl in a buffer containing 20 U RNasin (Promega), 10 mM Tris–HCl, pH 8.3, 1.5 mM MgCl₂, 50 mM KCl, 0.8 mM dNTPs, 200 ng oligodeoxynucleotide primer and about 7 ng FMDV RNA. For amplification with Pfu, the buffer recommended by the supplier was added to a final volume of 100 µl plus 200 ng of the second primer and 2.5 U of enzyme. The primers used contained restriction sites at their 5´ ends to facilitate cloning in appropriately digested pGEM4Z. To amplify the internal ribosome entry site (IRES) and the region between the two initiation AUG codons, the primer for cDNA synthesis was complementary to positions 1200–1183 of FMDV RNA and contained a SacI restriction site; the second primer for PCR amplification corresponded to nucleotides 569–587 of FMDV RNA and had a BamHI restriction site. To amplify the VP1-coding region, the primer for cDNA synthesis was complementary to positions 3888–3869 of FMDV RNA and contained a BamHI restriction site; the second primer for PCR amplification corresponded to nucleotides 3171–3192 of FMDV RNA and had a SacI restriction site. After treatment with phenol to inactivate the DNA polymerase, the DNA was recovered by ethanol precipitation (Sambrook et al., 1989). After digestion of both the PCR-amplified products and the vector pGEM4Z with SacI and BamHI, the DNA was electrophoresed through 1 % SeaPlaque agarose in 40 mM Tris–acetate, 1 mM EDTA, pH 8.0, the appropriate bands were excised from the gel and the DNA was purified by using the Gene Clean II kit as indicated by the manufacturer (Bio 101). After ligation overnight at 16 °C, the ligation product was transformed in E. coli DH-5 and transformants were isolated and analysed following standard procedures (Sambrook et al., 1989). Plasmid DNA was purified by using the Wizard Plus SV Minipreps kit (Promega). Two separate sets of molecular clones were obtained and analysed: 70 clones spanning the IRES and the region between the two AUG initiation codons and 70 clones corresponding to the VP1-coding region. Nucleotide sequencing was carried out in an ABI 373 automatic sequencer, as described previously (Escarmís et al., 1999).

Statistics. Standard statistical procedures used were those described in the package Hypothesis tests of the program Mathematica 3 (Wolfram Research).

Mutant spectrum complexity and composition during the process of fitness gain of a virus clone

Large population passages of FMDV C⁹₂₂ p0 led to an exponential increase in virus fitness (Escarmís et al., 1999) (Fig. 2). In order to quantify the genetic heterogeneity of the virus population during the process of fitness gain, 70 biological clones and 70 molecular clones from population C⁹₂₂ p50 were analysed by nucleotide sequencing. Three genomic regions were sequenced: residues 605–1038 (spanning the IRES), 1039–1122 (those between the two functional AUG initiation codons) and 3208–3834 (the capsid protein VP1-coding region) (Fig. 1). Nucleotide sequences of individual biological and molecular clones were compared with the consensus nucleotide sequence of C⁹₂₂ p50. The numbers and types of mutations did not reveal significant differences between the analysed biological and molecular clones. Two independent RT–PCR amplifications of the same template RNA population yielded molecular clones with similar distributions of mutations (Table 1). In addition to the standard (minimum) mutation frequency, calculated by counting repeated mutations only once, Table 1 also includes values for the maximum mutation frequency. The latter was calculated by considering the repeated mutations, and it has been used for some statistical evaluations of the mutant distributions (described in the next section and in the Discussion). In the mutant spectrum of C⁹₂₂ p50, the mean minimum mutation frequency found in the IRES element from the analysis of biological and molecular clones was 4.1x10^–4 substitutions per nucleotide. This value is 3.5- to 5.6-fold smaller than the mutation frequency found in the region between the two AUG codons (mean 1.8x10^–3 substitutions per nucleotide). This difference is expected, since the region between the two AUGs has been rendered non-functional with regard to the synthesis of Lab by U-1056 (Escarmís et al., 1996; Fig. 1). The mean minimum mutation frequency for the VP1-coding region was 5.3x10^–4 substitutions per nucleotide, only 1.3-fold larger than in the IRES. The ratio of transitions to transversions amounted to more than 29 in the IRES, 6 in the region between the two AUGs and 4 in the VP1-coding region. This abundance of transition over transversion mutations is expected from the misincorporation tendencies of RNA-dependent RNA polymerases (Domingo et al., 1978; Kuge et al., 1989; Schneider & Roossinck, 2000) and both biological and molecular clones reflected such a tendency (Table 1).

Table 1. Characterization of the mutant spectrum of C⁹₂₂ p50 quasispecies by analysis of biological and molecular clones

A total of 70 biological clones and 70 molecular clones was analysed; the molecular clones were derived from two independent RT–PCR amplifications (35 clones each). Procedures for the preparation and analysis of biological clones (BIO) and molecular clones (MOL) are described in Methods.

* Numbering of residues is according to Escarmís et al. (1996).
† Mutant residues are those that vary relative to the consensus nucleotide sequence population C⁹₂₂ p50. Numbers of mutations in MOL found in the two independent RT–PCR amplifications were: IRES, 9 and 6; between AUGs, 14 and 16; VP1-coding region, 20 and 14.
‡ The mutations found have been separated into Ts (transitions), Tv (transversions), Nsyn (non-synonymous, which lead to an amino acid replacement) and Syn (synonymous or silent, which do not lead to an amino acid replacement). The latter division does not apply to non-coding regions, as indicated by –.
§ The minimum mutation frequency is the number of different mutations found divided by the total number of nucleotides sequenced. The maximum mutation frequency is the total number of mutations found (as given in Table 2) divided by the total number of nucleotides sequenced. Mutation frequencies are expressed as substitutions per nucleotide.

Repeated mutations and mutational hot spots are identified in both biological and molecular clones

In order to compare the numbers and types of point mutations found in biological clones and in molecular clones derived from population C⁹₂₂ p50, all variant positions have been listed, together with the number of clones in which each mutation was found (Table 2). Of 25 point replacements that occurred more than once in either biological or molecular clones, 19 (76 %) were found both in biological and molecular clones. At four sites (genomic residues 1069, 1118, 3780 and 3801), the same mutation was found in six or more clones, and the mutation was a G A transition in all cases. Particularly striking was the occurrence of G-1118A in seven biological clones and 12 molecular clones. This mutational hot spot may be facilitated by U-1056, found in clone C⁹₂₂ and all its progeny populations (Escarmís et al., 1996, 1999) (Fig. 1) (see also Discussion).

The ²-test was applied to assess whether biological and molecular clones could be distinguished with regard to the numbers and types of mutations found in the two sets. No significant difference was found between biological clones and molecular clones in the number of times that a given mutation was represented (0.1<P<0.3; ², 1 degree of freedom). Equally indistinguishable were the distribution of mutations among the three genomic regions analysed (0.3<P<0.7; ², 2 degrees of freedom), the ratio of transition to transversion mutations (0.7<P<0.9; ², 4 degrees of freedom) and the ratio of synonymous to non-synonymous mutations (0.1<P<0.3; ², 1 degree of freedom). Student's t-test also indicated that the mean number of times that mutations were represented in the mutant spectrum was indistinguishable for biological clones and molecular clones (P=0.08). However, those mutations found three or more times among biological and molecular clones were slightly overrepresented in the latter (a total of 37 in molecular clones and 28 in biological clones). An exception was the case of C-886 U in the IRES; the mutation was found exclusively in three molecular clones (data in Table 2).

Insertions and deletions (indels) were found in the IRES and in the region between the two AUGs (Table 3). G-1118, the site of a mutational hot spot for G A transitions, was deleted in one molecular clone. A rich repertoire of adenylate insertions was found within positions 1119–1122, including one molecular clone with 20 additional adenylate residues, a length which is very close to the average present in the dominant sequence of the parental clone C⁹₂₂ (Escarmís et al., 1996).

Table 2. Point mutations in the mutant spectrum of C⁹₂₂ p50 identified by biological and molecular cloning

BIO, MOL refer to biological and molecular clones as defined in the Introduction. The numbers under BIO and MOL indicate the number of independent biological and molecular clones in which the mutation indicated was found. Procedures for preparation and analysis of biological and molecular clones are detailed in Methods. Nucleotide residues are numbered according to Escarmís et al. (1996). The location in the FMDV genome of the regions analysed is depicted in Fig. 1. Mutations in the IRES that are predicted to alter secondary structure or base pairings are underlined (see Discussion). Amino acid replacements are indicated in parentheses for non-synonymous mutations in the VP1-coding region. No amino acid replacements are indicated for genomic region 1039–1122 because a point deletion at residue 1056 rendered this region functionally irrelevant regarding the encoding of the Lab form of L (see Introduction). Amino acids are numbered according to their position in capsid protein VP1.

* These two mutations occurred in the same clones (see Discussion).
† This biological clone was heterogeneous and contained A and G at this position, with A dominant (about 70 % of the genomes, according to the peak pattern in the sequence obtained).

Clonal analysis of C⁹₂₂ p50 may also provide information on possible subsets of mutations present at different frequencies in the mutant spectrum. To approach this question, the probability that the point substitutions found in biological clones are also found in molecular clones (probability of coincidence) was calculated and compared with the actual experimental values (Table 2). For the IRES region (maximum mutation frequency of 4.75x10^–4 substitutions per nucleotide with 434 residues compared per genome; values taken from Table 1), the expected probability that any mutation found one, two, three, four or five times in one of the subsets of clones (biological or molecular) is also found one, two, three, four or five times in the other subset is 1.07x10^–3, a value 4.3-fold smaller than that found experimentally (data in Table 2). This overrepresentation of coincident mutations is significant (P=0.01), assuming that the probability of coincidence of mutations per nucleotide in biological and molecular clones follows a binomial distribution. For the VP1-coding region, the expected coincidences were 6.9–fold smaller than that found experimentally (P<0.0001), while in the region between the two AUGs there was no significant difference between the expected and actual number of mutations found in the two subsets of clones (P=0.36). Identical conclusions were obtained using the minimum mutation frequency values for the IRES (P=0.002) and VP1-coding region (P<0.0001), but not in the region between the two AUGs (P=0.003) (Table 1) (see Discussion). Thus, in the IRES and VP1-coding regions, some mutations from the mutant spectrum of C⁹₂₂ p50 occur more frequently than expected from the mean values of mutation frequency.

Point replacements, insertions and deletions may reflect quasispecies memory or be predictive of dominant genomes

Virus quasispecies may contain memory genomes in the form of minority components of the mutant spectra that reflect their past evolutionary history (Ruiz-Jarabo et al., 2000). The clonal analysis of C⁹₂₂ p50 reinforces the previous evidence of quasispecies memory and, furthermore, reveals that some mutations may anticipate a future, dominant sequence in quasispecies evolution. Specifically, A-3653 C, which results in the amino acid replacement K-149 T in VP1, was present in two biological clones and two molecular clones from C⁹₂₂ p50, and also in genomes that had been dominant at passages 0 and 20, and which became dominant again at passage 100 (Escarmís et al., 1999). Mutation C-3650 A, which results in the amino acid replacement T-148 K in VP1, was found in two biological clones and two molecular clones and it became dominant at passage 100 (Escarmís et al., 1999) (compare Fig. 2 and Table 2). The clonal analysis of C⁹₂₂ p50 extends the previous study on memory genomes harbouring an internal polyadenylate tract as a memory marker (Ruiz-Jarabo et al., 2000) by 70 additional clones. Despite their debilitating effect on virus fitness (Escarmís et al., 1999), additional adenylate residues were detected in 17 % of the clones analysed (Table 3). Thus, at a passage in which the virus population was actively gaining fitness, sequence analysis of individual clones revealed mutations that anticipated the future evolution of the population and other mutations that were remains of its evolutionary history.

The mutant spectrum of FMDV population C⁹₂₂ p50, in the exponential phase of fitness gain of a very debilitated virus (Fig. 2), has been analysed by nucleotide sequencing of 70 biological clones and 70 molecular clones. Three genomic regions have been sequenced, amounting to a total of 160160 nucleotides (Tables 1, 2 and 3). The results indicate that analyses of biological and molecular clones are equally valid experimental approaches to providing a representation of a virus quasispecies with regard to point substitutions, insertions and deletions. The initial studies that established the quasispecies structure and dynamics of RNA virus populations involved the sampling of nucleotide sequences and measurements of relative fitness of biological clones of a number of prokaryotic and eukaryotic viruses (Batschelet et al., 1976; Domingo et al., 1978; Holland et al., 1979; Spindler et al., 1982; for a review of the early work, see Domingo et al., 1988). However, the characterization of the mutant spectra of virus quasispecies is finding increasing application to the understanding of virus pathogenesis and evolution of viruses that either do not grow in cell culture or grow poorly (several human hepatitis viruses, the caliciviruses Norwalk virus and rabbit haemorrhagic disease virus, some enteric coronaviruses and papillomaviruses, among others; examples in Esteban et al., 1999; Forns et al., 1999; Flint et al., 2000; Domingo et al., 2001). For these systems, the characterization of mutant spectra may often necessitate the sequencing of molecular clones obtained after PCR or RT–PCR amplification of intracellular viral DNA or RNA. The conditions detailed in Methods using the high-fidelity Pfu DNA polymerase (Cline et al., 1996) provide a straightforward protocol to derive an accurate representation of point substitutions, indels and mutational hot spots. Under conditions of excess template (to avoid a molecular bottleneck in the amplification process), the reliability of the procedure is supported by the fact that the numbers and types of mutations and their distribution were statistically indistinguishable when biological clones and molecular clones were compared (Tables 1, 2 and 3). Although DNA polymerases of lower fidelity may also be appropriate when mutation frequencies in the target population are one or more orders of magnitude larger than in C⁹₂₂ p50 (examples in Nájera et al., 1995; Esteban et al., 1999; and references therein), the availability of increasing numbers of DNA polymerases with high fidelity and good processivity (Barnes, 1994; Cline et al., 1996) allows a reliable description of mutant spectra from the vast majority of populations of RNA and DNA viruses.

The genetic heterogeneity in the IRES of C⁹₂₂ p50 was comparable to that in the VP1-coding region (Table 1). Application of the M-fold program (included in the GCG package) indicated that 50 % of all point substitutions and the C insertion within residues 810–814 found in the IRES did not alter its predicted secondary structure, because these substitutions are located in loops or bulges (Pilipenko et al., 1989; Martínez-Salas et al., 1996). One of the molecular clones [MOL (IA)56; Table 4] included C-766 U and C-886 U, which are predicted to convert G:C into G:U base pairs in domains 3 and 4, respectively, with a total G of +3.8 kcal/mol (15.9 kJ/mol). The most drastic mutation was U-799 C, present in one biological clone and one molecular clone, which is predicted to disrupt a stem in a very conserved hammerhead structure present in domain 3 with G=+5.5 kcal/mol (23.0 kJ/mol) relative to the same IRES domain with the consensus nucleotide sequence. The remaining mutations are predicted to cause little modification in the secondary structure but various degrees of alteration of base pairing. The calculated differences in G never exceeded 2.8 kcal/mol (11.7 kJ/mol). Mutations that are not in predicted loops or bulges in the IRES are underlined in Table 2.

The statistical analysis of the distribution of mutations among biological and molecular clones suggests that some subsets of mutations are overrepresented in the mutant spectrum of C⁹₂₂ p50. This observation agrees with evidence from model studies with small RNA molecules amplified by Q replicase, which indicate that the mutant spectra were determined primarily by selection values rather than by mutation rates (Rohde et al., 1995; Biebricher, 1999). Interestingly, a difference between the expected and actual number of mutations present in biological and molecular clones was not observed in the region between the two AUGs, provided that maximum mutation frequencies (Table 1) were used in the statistical calculation. This is consistent with the facts that functional constraints have been largely relaxed by U-1056 (Escarmís et al., 1996) and that repeated mutations are the main contributors to blurring the difference between the observed and expected numbers of mutations. Either the virus replicase is more error-prone when copying some specific template residues or functional relaxation in the region between the two AUGs is non-uniform along the 84 positions. These results support the notion that virus quasispecies may enclose some sort of substructuring with regard to the abundance of subsets of mutants. This point is under further investigation.

Table 3. Insertions and deletions (indels) in the mutant spectrum of C⁹₂₂ p50 identified by biological and molecular cloning

BIO and MOL refer to biological and molecular clones. The numbers under BIO and MOL indicate the number of independent biological and molecular clones in which the indicated indel was found. Procedures for preparation and analysis of biological and molecular clones are detailed in Methods. Nucleotide residues are numbered according to Escarmís et al. (1999). The location in the FMDV genome of the regions analysed is depicted in Fig. 1. When the insertion of adenylate residues at positions 1119–1123 involved more than one additional adenylate, the number is given as a subscript; (A)_3–6 and (A)_8–13 mean that the numbers of additional adenylates in these two biological clones were heterogeneous.

A second question that the analysis of the mutant spectrum of C⁹₂₂ p50 allows to be addressed is the comparison between the expected and actual number of genomes harbouring one, two, three, four or five mutations (Table 4). Considering the analysis of biological clones for the IRES and the VP1-coding region (a length of 1061 nucleotides), the maximum mutation frequency is 6.5x10^–4 substitutions per nucleotide, which gives an expected mean number of mutations within these regions of 0.69 (6.5x10^–4x1061). The expected proportions of clones with no mutations or one, two, three, four or five mutations in the 1061 nucleotides analysed in biological clones are respectively 50, 35, 12, 2.7, 0.46 and 0.07 % (calculated according to the Poisson distribution P_K=m^Ke^–m/K!, where P_K is the probability of a genome having K mutations and m is the mean number of mutations per genome); the actual experimental values were respectively 50, 37, 10, 1.4, 0 and 1.4 %. The same calculation for the VP1-coding region among molecular and biological clones indicates that the predicted proportions of sequences with no mutations or one, two, three, four or five mutations are 62, 30, 7.3, 1.2, 0.14 and 0.01 %; the actual values were 64, 26, 7.9, 2.1, 0 and 0 %. In all cases, there is good agreement between the expected and actual distribution of mutations among the clones analysed (in both cases: 0.7<P<0.9; ², 1 degree of freedom), further supporting the conclusion that molecular and biological clones provided an indistinguishable representation of the C⁹₂₂ p50 quasispecies.

Analysis of the mutant spectrum of C⁹₂₂ p50 has detected the presence of memory genomes in virus quasispecies, documented previously in this and in another evolutionary lineage of FMDV (reviewed in Domingo, 2000; Ruiz-Jarabo et al., 2000). The memory markers identified were the replacement K-149 T in VP1 (Table 2) and the heterogeneous internal polyadenylate tract dominant in C⁹₂₂ p0, which contained an average of 19 additional adenylate residues (Escarmís et al., 1996) (Fig. 1). One molecular clone from C⁹₂₂ p50 included 20 additional adenylates (Table 3) and thus represents an accurate memory genome of the ancestral C⁹₂₂ p0 population, while other genomes with smaller numbers of adenylates must be regarded as derivatives of the founder genomes.

Additions or deletions of adenylate residues within positions 1119–1122, the hot-spot G-1118 A transition and the rare deletion of G-1118 can occur as a result of misalignment of the growing RNA strand or the template strand (Ripley, 1990) during viral RNA synthesis (Fig. 3). Misalignment mutagenesis events tend to occur in repeated sequences (Streisinger et al., 1966; Ripley, 1990; Denver et al., 2000; Funchain et al., 2000) and they may be favoured by the low stability of the poly(A)·(U) duplex. Homopolymeric poly(A)·poly(U) displays low melting temperature and a tendency to form a triple helix as the salt concentration is raised (Saenger, 1984; and references therein). Obviously, other molecular mechanisms are possible to account for the repertoire of point substitutions and indels in C⁹₂₂ p50 (Ripley, 1990). The presence of U-1056 prevents the expression of Lab and hence may explain both the higher mutation frequency observed in the region located between the two AUG initiation codons and the high frequency of the transition G-1118 A. This mutation would lead, in the absence of U-1056, to the replacement G-27 E in protease Lab. Since G-27 in Lab is conserved among European FMDV isolates (Ryan & Flint, 1997), its substitution by E may be deleterious when Lab is functional.

Fig. 3. Model of misalignment mutagenesis to explain insertions and deletions in the genomic region preceding the second functional AUG in the mutant spectrum of C⁹₂₂ p50. Sequences and residue numbers are from Escarmís et al. (1996, 1999). +, Plus-strand, genomic RNA (the second functional AUG is boxed); –, negative-strand, antigenomic RNA. (A) Misalignment of the nascent minus-strand may lead to genomes with an increased number of adenylate residues (underlined). (B) Misalignment of the template plus-strand may lead to genomes with a decreased number of adenylate residues. (C) Realignment of the misaligned nascent minus-strand may lead to mutation G-1118A. (D) Misalignment of the template plus-strand, once uridine residues in the minus strand have been incorporated up to position 1119, may lead to deletion of G-1118. Note that this is an unlikely event, due to the necessity of the misalignment occurring when the minus strand has grown to a specific position, while in (A)–(C), a number of alternative misalignment events during minus-strand synthesis can lead to insertion of A, deletions or the G-1118A transition. The schemes are based on mechanisms proposed previously and reviewed by Ripley (1990).

The mutant spectrum of C⁹₂₂ p50 also revealed two biological clones and two molecular clones with the replacement T-148 K, which predicted the dominance of this replacement in C⁹₂₂ p100 (Fig. 1; Table 2). The replacement T-148 K has previously been shown to occur in the course of passaging of FMDV C-S8c1 in BHK-21 cells (Díez et al., 1989; Borrego et al., 1993; Sevilla et al., 1996). Therefore, it could be argued that its presence in the mutant spectrum is fortuitous and unrelated to the dominance of this substitution in C⁹₂₂ p100. Although definitive proof that this replacement had a predictive value has not been obtained, other substitutions that have similarly been found upon passage of C-S8c1 in BHK-21 cells were not detected among the 140 clones analysed; such VP1 substitutions are Q-23 K, S-139 R, A-140 V, L-144 V, A-145 P, T-148 A, T-149 A, T-149 K and T-150 K (Díez et al., 1989; Borrego et al., 1993; Sevilla et al., 1996). None of these replacements, with the exception of T-148 K and T-148 A (which are mutually exclusive as dominant in an FMDV population), was represented among the 13 amino acid replacements in VP1 scored in the mutant spectrum of C⁹₂₂ p50 (Table 2). The biological clones BIO 12 and BIO 31 (Table 4) included C-3650 A (T-148 K) and A-3653 C (K-149 T) and predicted the dominance of K-148 and T-149 in C⁹₂₂ p100. The same biological clones, however, did not include K-150, which also became dominant in C⁹₂₂ p100. Two consecutive lysine residues have never been observed within VP1 residues 148–150 in consensus sequences or in individual clones, even after strong selection with polyclonal antibodies directed specifically to this VP1 loop (Borrego et al., 1993). Such a K duplex in a helical region of the G–H loop of VP1 is likely to be deleterious for integrin recognition (Mason et al., 1994; Berinstein et al., 1995; Verdaguer et al., 1995). Therefore, it is likely that some components of a mutant spectrum may have a predictive value of those genomes that will become dominant later in the evolutionary lineage. This interesting possibility will be investigated further.

Table 4. Biological and molecular clones from the mutant spectrum of C⁹₂₂ p50 that contained more than one mutation

BIO and MOL indicate biological and molecular clones. Because the IRES and the region between the two AUGs were cloned in E. coli independently of the VP1-coding region, clones with two or more mutations had to be counted separately; this is indicated by MOL (IA) (molecular clones of the IRES and the region between the two AUGs) and MOL (VP) (molecular clones of the VP1-coding region). A total of 70 biological clones, 70 MOL (IA) clones and 70 MOL (VP) clones was analysed. Procedures are detailed in Methods. Residues are numbered according to Escarmís et al. (1996). All mutations found in the clones analysed are listed in Table 2.

In conclusion, a reliable characterization of the mutant spectrum of a virus quasispecies at the nucleotide sequence level can be obtained through the analysis of biological clones or molecular clones. The results revealed the great complexity of a mutant spectrum of a clonal population in the process of fitness gain in a constant biological environment. The distribution of mutations provided evidence of substructuring within the mutant spectrum of the C⁹₂₂ p50 quasispecies. Some mutants reflected the past evolutionary history of the population, while other mutants anticipated those that will become dominant at a later stage of the evolutionary process. This observation may be of practical relevance, in that the quantification of mutations related to variations in B cell or T cell epitopes and to resistance to antiviral agents, which may be present in different proportions in the mutant spectrum, may guide decisions on alternative immunotherapeutic or antiviral regimens. There is increasing evidence that fitness values, virus load and quasispecies complexity may be relevant to the pathogenic potential of viruses (Rowe et al., 1997; Farci et al., 2000; Quiñones-Mateu et al., 2000) and to the response of an infected host to antiviral treatment (Pawlotsky et al., 1998). The present study encourages quasispecies composition analyses at the nucleotide sequence level for diagnostic and therapeutic purposes.

We are indebted to J. Perez-Mercader for support, to M. Dávila and G. Gómez-Mariano for expert technical assistance and to F. J. Doblas-Reyes for help with statistical procedures. Work at the CBMSO was supported by grants FIS 98/0054-01 and PM 97-0060-C02-01 and Fundación R. Areces. Work at CAB was supported by the EU and INTA. A.A. was supported by fellowships from UAM and CAM.

Barnes, W. M. (1994). PCR amplification of up to 35-kb DNA with high fidelity and high yield from lambda bacteriophage templates. Proceedings of the National Academy of Sciences, USA 91, 2216–2220.

Batschelet, E., Domingo, E. & Weissmann, C. (1976). The proportion of revertant and mutant phage in a growing population, as a function of mutation and growth rate. Gene 1, 27–32.

Belsham, G. J. (1993). Distinctive features of foot-and-mouth disease virus, a member of the picornavirus family; aspects of virus protein synthesis, protein processing and structure. Progress in Biophysics & Molecular Biology 60, 241–260.

Berinstein, A., Roivainen, M., Hovi, T., Mason, P. W. & Baxt, B. (1995). Antibodies to the vitronectin receptor (integrin _{V 3}) inhibit binding and infection of foot-and-mouth disease virus to cultured cells. Journal of Virology 69, 2664–2666.

Biebricher, C. K. (1999). Mutation, competition and selection as measured with small RNA molecules. In Origin and Evolution of Viruses, pp. 65–85. Edited by E. Domingo, R. G. Webster & J. J. Holland. San Diego: Academic Press.

Borrego, B., Novella, I. S., Giralt, E., Andreu, D. & Domingo, E. (1993). Distinct repertoire of antigenic variants of foot-and-mouth disease virus in the presence or absence of immune selection. Journal of Virology 67, 6071–6079.

Chao, L. (1990). Fitness of RNA virus decreased by Muller's ratchet. Nature 348, 454–455.

Cline, J., Braman, J. C. & Hogrefe, H. H. (1996). PCR fidelity of pfu DNA polymerase and other thermostable DNA polymerases. Nucleic Acids Research 24, 3546–3551.

de la Torre, J. C., Martínez-Salas, E., Diez, J., Villaverde, A., Gebauer, F., Rocha, E., Dávila, M. & Domingo, E. (1988). Coevolution of cells and viruses in a persistent infection of foot-and-mouth disease virus in cell culture. Journal of Virology 62, 2050–2058.

Denver, D. R., Morris, K., Lynch, M., Vassilieva, L. L. & Thomas, W. K. (2000). High direct estimate of the mutation rate in the mitochondrial genome of Caenorhabditis elegans. Science 289, 2342–2344.

Díez, J., Mateu, M. G. & Domingo, E. (1989). Selection of antigenic variants of foot-and-mouth disease virus in the absence of antibodies, as revealed by an in situ assay. Journal of General Virology 70, 3281–3289.

Domingo, E. (2000). Viruses at the edge of adaptation. Virology 270, 251–253.

Domingo, E., Sabo, D., Taniguchi, T. & Weissmann, C. (1978). Nucleotide sequence heterogeneity of an RNA phage population. Cell 13, 735–744.

Domingo, E., Holland, J. J. & Ahlquist, P. (editors) (1988). RNA Genetics. Boca Raton, FL: CRC Press.

Domingo, E., Biebricher, C., Eigen, M. & Holland, J. J. (2001). Quasispecies and RNA Virus Evolution: Principles and Consequences. Austin, TX: Landes Bioscience.

Drake, J. W. & Holland, J. J. (1999). Mutation rates among RNA viruses. Proceedings of the National Academy of Sciences, USA 96, 13910–13913.

Duarte, E., Clarke, D., Moya, A., Domingo, E. & Holland, J. (1992). Rapid fitness losses in mammalian RNA virus clones due to Muller's ratchet. Proceedings of the National Academy of Sciences, USA 89, 6015–6019.

Eigen, M. (1971). Self-organization of matter and the evolution of biological macromolecules. Naturwissenschaften 58, 465–523.

Eigen, M. & Biebricher, C. K. (1988). Sequence space and quasispecies distribution. In RNA Genetics, pp. 211–245. Edited by E. Domingo, P. Ahlquist & J. J. Holland. Boca Raton, FL: CRC Press.

Escarmís, C., Dávila, M., Charpentier, N., Bracho, A., Moya, A. & Domingo, E. (1996). Genetic lesions associated with Muller's ratchet in an RNA virus. Journal of Molecular Biology 264, 255–267.

Escarmís, C., Dávila, M. & Domingo, E. (1999). Multiple molecular pathways for fitness recovery of an RNA virus debilitated by operation of Muller's ratchet. Journal of Molecular Biology 285, 495–505.

Esteban, J. I., Martell, M., Carman, W. F. & Gomez, J. (1999). The impact of rapid evolution of the hepatitis viruses. In Origin and Evolution of Viruses, pp. 345–376. Edited by E. Domingo, R. G. Webster & J. J. Holland. San Diego: Academic Press.

Farci, P., Shimoda, A., Coiana, A., Diaz, G., Peddis, G., Melpolder, J. C., Strazzera, A., Chien, D. Y., Munoz, S. J., Balestrieri, A., Purcell, R. H. & Alter, H. J. (2000). The outcome of acute hepatitis C predicted by the evolution of the viral quasispecies. Science 288, 339–344.

Flint, S. J., Enquist, L. W., Krug, R. M., Racaniello, V. R. & Skalka, A. M. (2000). Virology. Molecular Biology, Pathogenesis and Control. Washington, DC: American Society for Microbiology.

Forns, X., Purcell, R. H. & Bukh, J. (1999). Quasispecies in viral persistence and pathogenesis of hepatitis C virus. Trends in Microbiology 7, 402–410.

Funchain, P., Yeung, A., Stewart, J. L., Lin, R., Slupska, M. M. & Miller, J. H. (2000). The consequences of growth of a mutator strain of Escherichia coli as measured by loss of function among multiple gene targets and loss of fitness. Genetics 154, 959–970.

Holland, J. J., Grabau, E. A., Jones, C. L. & Semler, B. L. (1979). Evolution of multiple genome mutations during long-term persistent infection by vesicular stomatitis virus. Cell 16, 495–504.

Holland, J. J., de la Torre, J. C., Clarke, D. K. & Duarte, E. (1991). Quantitation of relative fitness and great adaptability of clonal populations of RNA viruses. Journal of Virology 65, 2960–2967.

Holland, J. J., de la Torre, J. C. & Steinhauer, D. A. (1992). RNA virus populations as quasispecies. Current Topics in Microbiology and Immunology 176, 1–20.

Kuge, S., Kawamura, N. & Nomoto, A. (1989). Strong inclination toward transition mutation in nucleotide substitutions by poliovirus replicase. Journal of Molecular Biology 207, 175–182.

Martínez-Salas, E., Regalado, M. P. & Domingo, E. (1996). Identification of an essential region for internal initiation of translation in the aphthovirus internal ribosome entry site and implications for viral evolution. Journal of Virology 70, 992–998.

Mason, P. W., Rieder, E. & Baxt, B. (1994). RGD sequence of foot-and-mouth disease virus is essential for infecting cells via the natural receptor but can be bypassed by an antibody-dependent enhancement pathway. Proceedings of the National Academy of Sciences, USA 91, 1932–1936.

Nájera, I., Holguín, A., Quiñones-Mateu, M. E., Muñoz-Fernández, M. A., Nájera, R., López-Galíndez, C. & Domingo, E. (1995). Pol gene quasispecies of human immunodeficiency virus: mutations associated with drug resistance in virus from patients undergoing no drug therapy. Journal of Virology 69, 23–31.

Novella, I. S., Clarke, D. K., Quer, J., Duarte, E. A., Lee, C. H., Weaver, S. C., Elena, S. F., Moya, A., Domingo, E. & Holland, J. J. (1995 a). Extreme fitness differences in mammalian and insect hosts after continuous replication of vesicular stomatitis virus in sandfly cells. Journal of Virology 69, 6805–6809.

Novella, I. S., Duarte, E. A., Elena, S. F., Moya, A., Domingo, E. & Holland, J. J. (1995 b). Exponential increases of RNA virus fitness during large population transmissions. Proceedings of the National Academy of Sciences, USA 92, 5841–5844.

Novella, I. S., Elena, S. F., Moya, A., Domingo, E. & Holland, J. J. (1995 c). Size of genetic bottlenecks leading to virus fitness loss is determined by mean initial population fitness. Journal of Virology 69, 2869–2872.

Nowak, M. A. (1992). What is a quasispecies? Trends In Ecology & Evolution 4, 118–121.

Pawlotsky, J. M., Germanidis, G., Neumann, A. U., Pellerin, M., Frainais, P. O. & Dhumeaux, D. (1998). Interferon resistance of hepatitis C virus genotype 1b: relationship to nonstructural 5A gene quasispecies mutations. Journal of Virology 72, 2795–2805.

Pilipenko, E. V., Blinov, V. M., Chernov, B. K., Dmitrieva, T. M. & Agol, V. I. (1989). Conservation of the secondary structure elements of the 5´-untranslated region of cardio- and aphthovirus RNAs. Nucleic Acids Research 17, 5701–5711.

Quiñones-Mateu, M. E., Ball, S. C., Marozsan, A. J., Torre, V. S., Albright, J. L., Vanham, G., van der Groen, G., Colebunders, R. L. & Arts, E. J. (2000). A dual infection/competition assay shows a correlation between ex vivo human immunodeficiency virus type 1 fitness and disease progression. Journal of Virology 74, 9222–9233.

Ripley, L. S. (1990). Frameshift mutation: determinants of specificity. Annual Review of Genetics 24, 189–213.

Rohde, N., Daum, H. & Biebricher, C. K. (1995). The mutant distribution of an RNA species replicated by Q replicase. Journal of Molecular Biology 249, 754–762.

Rowe, C. L., Baker, S. C., Nathan, M. J. & Fleming, J. O. (1997). Evolution of mouse hepatitis virus: detection and characterization of spike deletion variants during persistent infection. Journal of Virology 71, 2959–2969.

Rueckert, R. R. (1996). Picornaviridae: the viruses and their replication. In Fields Virology, 3rd edn, pp. 609–654. Edited by B. N. Fields, D. M. Knipe & P. M. Howley. Philadelphia: Lippincott–Raven.

Ruiz-Jarabo, C. M., Arias, A., Baranowski, E., Escarmís, C. & Domingo, E. (2000). Memory in viral quasispecies. Journal of Virology 74, 3543–3547.

Ryan, M. D. & Flint, M. (1997). Virus-encoded proteinases of the picornavirus super-group. Journal of General Virology 78, 699–723.

Saenger, W. (1984). Principles of Nucleic Acid Structure. New York: Springer.

Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning: A Laboratory Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory.

Schneider, W. L. & Roossinck, M. J. (2000). Evolutionarily related Sindbis-like plant viruses maintain different levels of population diversity in a common host. Journal of Virology 74, 3130–3134.

Sevilla, N., Verdaguer, N. & Domingo, E. (1996). Antigenically profound amino acid substitutions occur during large population passages of foot-and-mouth disease virus. Virology 225, 400–405.

Sobrino, F., Dávila, M., Ortín, J. & Domingo, E. (1983). Multiple genetic variants arise in the course of replication of foot-and-mouth disease virus in cell culture. Virology 128, 310–318.

Sobrino, F., Sáiz, M., Jiménez-Clavero, M. A., Núñez, J. I., Rosas, M. F., Baranowski, E. & Ley, V. (2001). Foot-and-mouth disease virus: a long known virus, but a current threat. Veterinary Research 32, 1–30.

Spindler, K. R., Horodyski, F. M. & Holland, J. J. (1982). High multiplicities of infection favor rapid and random evolution of vesicular stomatitis virus. Virology 119, 96–108.

Streisinger, G., Okada, Y., Emrich, J., Newton, J., Tsugita, A. & Terzaghi, E. (1966). Frameshift mutations and the genetic code. Cold Spring Harbor Symposia on Quantitative Biology 31, 77–84.

Toja, M., Escarmis, C. & Domingo, E. (1999). Genomic nucleotide sequence of a foot-and-mouth disease virus clone and its persistent derivatives. Implications for the evolution of viral quasispecies during a persistent infection. Virus Research 64, 161–171.

Verdaguer, N., Mateu, M. G., Andreu, D., Giralt, E., Domingo, E. & Fita, I. (1995). Structure of the major antigenic loop of foot-and-mouth disease virus complexed with a neutralizing antibody: direct involvement of the Arg–Gly–Asp motif in the interaction. EMBO Journal 14, 1690–1696.

Weaver, S. C., Brault, A. C., Kang, W. & Holland, J. J. (1999). Genetic and fitness changes accompanying adaptation of an arbovirus to vertebrate and invertebrate cells. Journal of Virology 73, 4316–4326.

Yuste, E., Sánchez-Palomino, S., Casado, C., Domingo, E. & López-Galíndez, C. (1999). Drastic fitness loss in human immunodeficiency virus type 1 upon serial bottleneck events. Journal of Virology 73, 2745–2751.

JGV Direct table of contents

© 2001 SGM

This article is now available in the May 2001 print issue of JGV (vol. 82, 1049–1060). The complete issue of the journal may be seen in electronic form on JGV Online.