Michael B.A. Oldstone wrote that “LCMV has proven to be a Rosetta stone for uncovering numerous phenomena in virology and immunology”. The several crucial achievements were listed by Oldstone in an introduction to an outstanding compilation of review articles on arenavirus biology [1]. Yet, arenaviruses pose important challenges regarding the mechanism of replication and regulation of gene expression of a segmented genome, as well as disease mechanisms, prevention and treatment. Persistent arenavirus infections are highly prevalent among rodents, and when some of the viruses infect humans they may cause severe disease, including hemorrhagic fevers [2,3,4]. The prototypic arenavirus LCMV is increasingly regarded as a neglected human pathogen [5,6,7,8]. Unfortunately, the number of preventive or therapeutic possibilities for arenavirus-associated disease is very restricted. Despite promising developments [9,10,11,12], no licensed vaccines are generally available, and current therapy is essentially limited to an off-label use of the antiviral agent ribavirin (1-β-D-ribofuranosyl-1-H-1,2,4-triazole-3-carboxamide) (Rib) [13,14]. This is why arenavirus-related diseases are an interesting target to explore new antiviral designs.

Lack of effective vaccines and antiviral treatments is a general difficulty for diseases associated with highly variable RNA viruses, due to quasispecies dynamics [15]. The root of the problem is that mutations occur at such a high rate during RNA genome replication that even what is generally termed the “wild type” virus is, at any time, an average of a multitude of sequences. Although our attention is often fixed on a consensus sequence or on specific mutants identified in mutant spectra, in reality we deal with mutant clouds of rather indeterminate composition despite transient dominance of some specific mutant residues. Analyses of mutant spectra by ultra-deep sequencing have confirmed the extreme complexity of viral populations in nature, in support of previous evidence obtained by molecular cloning and sequencing ([16,17], among other examples [15]). LCMV participates of high mutation rates and quasispecies dynamics, with a direct implication in viral persistence and disease ([18,19,20,21,22,23,24,25,26,27,28,29,30,31,32]; reviews in [33,34]). The range of mutation frequencies calculated with genomes sampled from mutant spectra of LCMV populations passaged in cell culture in absence of mutagenic agents is in the range of 1.0 × 10^-4 to 7.5 × 10^-4 substitutions per nucleotide (s × nt^-1), and mutation frequencies reached 3.6 × 10^-3 and 1.1 × 10^-3 s × nt^-1 when the virus was passaged in the presence of 5-fluorouracil (FU) or Rib, respectively [35,36,37,38,39,40]. Mutation frequencies are in agreement with values of Shannon entropy, a measure of the proportion of different sequences in a genome distribution. Both parameters are used to quantify the complexity of mutant spectra, which is influenced by the number of passages from a clonal origin (i.e. starting with a plaque-purified virus), the multiplicity of infection (MOI), and the viral gene analyzed. Several studies indicate that the complexities of mutant spectra of LCMV are comparable to those determined for other RNA viruses [15].

Mutations often entail a fitness cost, and this applies also to drug-escape mutants. The almost systematic selection of drug-resistant viral mutants can be explained by the existence of multiple pathways to resistance, some with limited fitness cost (under a genomic sequence context or a given environment), and to the selection of compensatory (fitness-enhancing) mutations in viral genomes that continue expressing the resistance phenotype [15]. Under this scenario, it has been amply recognized that new antiviral strategies must be investigated to avoid or minimize the selection of drug or multi-drug-escape mutants, and the consequent treatment failure. Strategies already implemented or under development are combination therapies, splitting of treatment between an induction and a maintenance regimen, use of drugs that target cellular functions, and combination of immunotherapy and chemotherapy. As reviewed in [15], none of these strategies is free of the problem of selection of escape mutants.

An additional strategy now under investigation is lethal mutagenesis, also termed virus entry into error catastrophe, in recognition of its conceptual origins. It consists in achieving large reductions of viral load and ideally virus extinction by increasing the mutation rate of the virus above a critical error rate that sets a limit for virus viability. Current research involves the use of mutagenic base or nucleoside analogues which are converted intracellularly in the nucleoside-triphosphate derivatives which are incorporated into viral RNA during elongation. They do not act as chain terminators, a mechanism that results mainly in inhibition, but they allow continued chain elongation, a mechanism that results in mutagenesis. The latter is a result of the ambiguous pairing with the corresponding complementary purine or pyrimidine nucleotides, when the analogue is either a substrate or a template residue [41]. Lethal mutagenesis was inspired in the concept of crossing an error threshold, or transition into error catastrophe, established as an important corollary of quasispecies theory [42,43]. Crossing the error threshold implies loss of genetic information. It was reasoned that drugs that increase the mutation rate of a virus above a viability threshold may highly diminish the chances of virus escape. Mutagenesis can provide an advantage over conventional non-mutagenic inhibitors because mutagenized genomes may suppress the most infective subpopulations of genomes coexisting in the same quasispecies, as discussed in the next section. A number of base and nucleoside analogues are currently under development as potential new antiviral mutagenic agents [44,45,46,47,48,49,50].

At present it is not known whether in the context of a clinical application of lethal mutagenesis [51], mutagen-resistant mutants will be as readily selected as inhibitor-resistant mutants. Rib-resistant mutants have been isolated in cell culture [52,53,54,55,56,57,58,59,60,61,62,63,64,65], and in patients subjected to Rib monotherapy [66]. Rib resistance can have multiple mechanisms, as evidenced by the fact that resistance mutations have been mapped in several viral genes. When resistance is associated with amino acid substitutions in the polymerase, it can be due to restoration of polymerase activity or to alterations of nucleotide recognition. The latter can occur through at least three mechanisms: average increase of template-copying fidelity, specific decrease of Rib-triphosphate incorporation (without a significant effect on general fidelity), or modulation of the transition types produced in the presence of Rib (without alteration of the average mutation frequency among the components of the mutant spectrum) [55,57,58,59,61].

LCMV has had an important contribution regarding the feasibility of a lethal mutagenesis-based approach in vivo [37], and in the proposal of the lethal defection model of virus extinction (participation of defective viruses in decrease of replicative competence of a viral quasispecies [67]), two of the highlights of lethal mutagenesis research (Table 1). In addition, early experiments evidenced that LCMV could be efficiently extinguished by enhanced mutagenesis [35] (comment by Eigen [68]). The transition towards extinction did not involve any modification of the consensus genomic sequence, and entailed a 10²- to 10³-fold decrease in specific infectivity of LCMV [36]. These were revealing observations to interpret the molecular events associated with virus extinction.

Lethal defection introduced a role of internal interactions among components of a mutant spectrum in the process of virus extinction. The key observation was that when a persistent LCMV infection of BHK-21 cells was perturbed by addition of the mutagen FU, viral infectivity decreased at a faster rate than the level of viral RNA [67]. The unexpectedly faster kinetics of loss of infectivity than viral RNA was interpreted as due to the presence of a class of defective genomes, termed “defectors”, that were generated as a consequence of mutagenesis. Defectors could replicate their RNA but jeopardize formation of infectious particles, contributing to a decline of infectivity that was more substantial than the decline of viral RNA. The experimental findings were supported by in silico simulations of the outcome of LCMV replication in the absence or presence of defectors under different mutagenic intensities [67,77]. The combination of theory and experiment led to the proposal of the lethal defection model, according to which defectors play an important role in virus extinction. The term defector had been previously used to refer to other types of non-functional genomes in models of RNA virus evolution [78,79]. In lethal mutagenesis, a defector is a genome that manifests some defect during its replication cycle, that may or may not complete production of infectious particles, but that is competent in RNA replication. The latter feature is essential to express an interfering activity against fully infectious viral genomes, as documented with specific foot-and-mouth disease (FMDV) capsid and polymerase mutants displaying high or low levels or RNA replication [80]. Furthermore, an interfering, replication-competent virus with two polymerase substitutions lost its interfering activity when a third polymerase mutation that abolished RNA replication was introduced in the genome [80]. Current models suggest that interference can come about through expression of non-functional or suboptimal viral proteins by the defectors, since a majority of viral proteins are multi-functional and active through formation of homomeric or heteromeric complexes with viral or host proteins (for example, proteins that must interact to form the viral capsid or RNA replication complexes). A protein that includes an amino acid substitution at critical contact site, and that is encoded by a subset of genomes from a mutant spectrum, may contribute to formation of either non-functional or suboptimal complexes, thereby decreasing the replicative efficiency of the viral genomes that exploit that particular protein complex. In a mutagenized mutant spectrum, this mechanism may have a multiplicative effect as a result of the action of many low frequency variants harboring “defector-prone” mutations that may adversely affect one or more viral functions. This view is an extension of interference as defined in classic genetics [81], applied to a context of multitudes of micro-interference events in a mutant cloud [15].

A direct biochemical evidence of formation of suboptimal protein complexes as a result of mutagenesis is still lacking. However, interference by mutant spectra has been reported for a variety of virus-host systems in cell culture and in vivo, and suppression can limit replication of high fitness, virulent or drug-resistant mutants [32,67,82,83,84,85]. Internal interactions within mutant spectra confer viral quasispecies an identity as units of selection, a hallmark of the quasispecies concept [42]. In the original quasispecies theory the unit of selection was founded on the cross-talk through mutation among components of the same mutant spectrum. In viruses the cross-talk occurs through interactions among expression products. Such interactions represent an important conceptual departure over classic models that viewed quasispecies (or “intra-host variations”, as sometimes inaccurately named) as mere aggregates of mutants resulting from a mutation-selection equilibrium [15,86]. Both, in theoretical quasispecies and in viral quasispecies an interacting mutant ensemble is the unit of selection.

Quasispecies did not exist as a scientific concept until 1971, when Manfred Eigen published his pioneer treatise [102]. Since its completion by Eigen and Schuster [42], the theory has been extended to finite populations of replicons under non-equilibrium conditions, and several of its ground-breaking predictions (adaptability, limitations for sustained maintenance of information, ensembles of mutants as units of selection, etc.) have been applied and confirmed with complex biological entities such as the RNA viruses, with their highly compact genomes. Despite many unknowns in viral dynamics, the new insight provided by quasispecies is now permeating several aspects of RNA virus biology, including medical aspects such as the design of antiviral therapies. Treatment of arenavirus infections may also benefit from these developments.