Quasispecies and the implications for virus persistence and escape

Abstract

Background: In the 1970s Manfred Eigen and colleagues proposed a new model of molecular evolution to explain adaptability and rapid evolution of simple replicons, as those that probably populated the earth at the onset of life. This model of evolution placed emphasis on mutant generation, to the point of invalidating the concept of wild-type genomes as a defined sequence of nucleotides. In striking similarity with the proposals for such early replicons, present-day RNA viruses consist of complex distributions of nonidentical but closely related genomes termed quasispecies.

Objectives: To discuss indeterminations inherent to a quasispecies structure and to the analytical procedures to define it, biological implications of quasispecies, and the need to take into account this type of population structure, in order to design effective strategies to prevent and control diseases caused by highly variable viruses.

Results: Quasispecies have many biological implications, extending from viral pathogenesis to the emergence of new pathogens, rapid antigenic variation, and alterations in cell tropism, virulence, host range and viral gene expression.

Conclusions: Diseases caused by highly variable RNA viruses prove very difficult to control and vaccine development against such viruses are largely unsuccessful. It is important to understand quasispecies composition and dynamics, as quasispecies are an important step in the natural history of RNA viruses.

1. Introduction

Quasispecies defines a genetic organisation of simple replicons at the population level, first proposed by Eigen, Schuster and their colleagues (Eigen, 1971, Eigen and Schuster, 1979). It means that mutant genomes are generated at such high rates that each individual genomic molecule differs on average in one or a few nucleotides from the other molecules, and each genomic sequence may deviate from the average or consensus sequence. For example, in the proposal of the quasispecies model for phage Qβ it was found that the most abundant genome amounted to 10–20% of the population. This experimental finding agreed with theoretical predictions of Eigen and colleagues, given the complexity of the Qβ genome and the mutation rate during its replication (Domingo et al., 1978).

Since this first proposal that bacteriophage Qβ, replicating in Escherichia coli, displayed features of a quasispecies, a steadily increasing number of RNA viruses have been shown to share such a population structure (Holland et al., 1982, Domingo et al., 1997). However, there has been a slow realisation of the biological implications of quasispecies. Since adaptability of RNA viruses is key to viral pathogenesis and strategies for disease control, it would seem obvious that quasispecies should have been regarded as highly relevant to these questions (Domingo, 1989, Domingo, 1996, Domingo and Holland, 1992, Duarte et al., 1994, Novella et al., 1995). This paper updates some observations on quasispecies, and highlights a few critical issues.

2. Trying to cope with indetermination

The main departure of quasispecies from previous models of molecular evolution is the consideration of the wild-type not as a genome with a single nucleotide sequence, but as a distribution of genomes which act as a unit of selection. The absence of a defined wild-type has been confirmed experimentally (Domingo et al., 1978, Domingo et al., 1995, Domingo et al., 1997). A viral genome is statistically defined but individually indeterminate. It is not possible to attribute a biological behaviour to a genomic nucleotide sequence with complete certainty. We must, therefore, rely on the statistical accumulation of data. In a study of antigenic sites of foot-and-mouth disease virus (FMDV) (the example applies to any other virus) the assignment of a neutralising monoclonal antibody to a given epitope necessitates that different clonal preparations of FMDV repeatedly produce escape mutants with amino acid substitutions at the same epitope region (Martínez et al., 1997).

Since it is not possible at present to sequence an individual genomic molecule, a large number of genomes (10⁸–10¹¹) must be gathered, following either biological or molecular cloning, before applying the chemistry for sequence determination. Biological cloning may imply some selection bias on viruses. This led to the timely statement concerning human immunodeficiency virus (HIV)-1 quasispecies analysis “to culture is to disturb” (Meyerhans et al., 1989). Other variations may be found in the mutant spectra derived from reverse transcription (RT) polymerase chain reaction (PCR) amplification procedures, or sequence determination from molecular clones obtained in E. coli or other cell-vector systems. Variations and biases may arise from the selective cloning of subsets of genomes, or other mechanisms (Szybalski, 1993, Forns et al., 1997) and cloning in alternative vector-cell systems may be required. RT-PCR errors in amplified genomes must be controlled by using the same protocols for experimental and problem samples (Nájera et al., 1995). It is not surprising, therefore, that the level of genetic heterogeneity of viruses cultured in cells is not significantly different from those estimated using biological molecular cloning or from RT-PCR. Molecular error-prone replication is based on the absence of proofreading-repair, 3′–5′ exonuclease activity (Steinhauer et al., 1992, Sousa, 1996). This applies to the enzymes that replicate or retrotranscribe viral genomes as well as to those that amplify their genomes in vitro often with comparable numbers of replication rounds. Heterogeneities are observed when visualising a consensus sequence, i.e. double bands or similar. In these cases, populations almost systematically segregate into the expected distinct genomes when biological or molecular cloning is undertaken. This finding may be statistically predicted as only a minority of mutations or heterogeneous positions encountered in handling viral genomes will actually have an effect on the result of the biological or cloning assays.

The primary conclusion from published data is that RNA viruses (including viruses which use RNA as replicative intermediate such as hepatitis B virus) behave largely as predicted by the quasispecies model of molecular evolution. Quasispecies swarms, and not any sort of defined genetic entity, are those that replicate in infected hosts as they produce disease (Holland et al., 1992 Domingo et al., 1995 Marcus et al., 1998).

3. Biological implications of the quasispecies

Virologists use an extended definition of quasispecies so that recombination can also contribute to the mutant spectrum. Rowe et al. (1997)showed that the complexity of the coronavirus murine hepatitis virus quasispecies, defined by the presence of recombinant genomes and point mutations, influences the pathogenic potential of this virus in mice. In the extended definition, useful to virology, quasispecies are dynamic distributions of non-identical but closely-related mutant and recombinant viral genomes. They are subjected to a continuous process of genetic variation, competition and selection, and act as a unit of selection. The links between the original theoretical quasispecies concept and viral quasispecies have been discussed in several recent papers (Domingo et al., 1995, Eigen, 1996, Domingo and Holland, 1997).

If Eigen’s model had not been formulated it would be difficult to reconcile many observations on genotypic and phenotypic variations of RNA viruses. Biological implications of quasispecies extend from viral pathogenesis to the emergence of new viral pathogens (Morse, 1993). They include rapid antigenic variation (antibody-escape and cytotoxic T-lymphocyte (CTL)-escape), alterations in cell tropism, virulence, host range and viral gene expression (Domingo, 1989, Domingo, 1996, Jackson et al., 1996, Domingo et al., 1997, Sa-Carvalho et al., 1997) (Table 1 ).

Table 1.

Recent examples of biological implications of quasispecies

Virus	Observation	References
Foot-and-mouth disease virus	Altered cell tropism of the minority of the mutant spectrum	Jackson et al., 1996, Sa-Carvalho et al., 1997
	Dispensability of essential motif	Martínez et al. (1997)
Vesicular stomatitis virus	Different ability of viral clones to induce interferon	Marcus et al. (1998)
Feline calicivirus	Evidence of antigenic evolution in vivo	Radford et al. (1998)
Hepatitis C virus	Response to IFN related to quasispecies complexity	Le Guen et al. (1997); Pawlotsky et al. (1998, unpublished data)

Additional examples can be found in Domingo (1996)and Domingo et al. (1997).

4. Types, subtypes and beyond

One of the most obvious consequences of quasispecies structure is the genetic and antigenic diversification of RNA viruses in nature. This has triggered efforts to classify any viral pathogen into different types or subtypes. While classification has clear advantages for diagnostic, clinical and epidemiological work, it tends to assign, to particular viral groups, some biological properties unlikely to be valid as additional isolates are included in the collections. An example is the subtyping of FMDV. This was stopped two decades ago when, after having isolated a representative of subtype 65, it was realised that virtually any isolate could be classified as a new subtype. It was found that additional antibodies allowed increased discrimination among isolates. Viral types and subtypes are just the ‘tip of the iceberg’ of a more profound and fundamental phenomenon: the continuous dynamics of mutant generation, competition, selection and random events which push viral quasispecies towards diversification.

In some cases serological classifications (and possibly phylogenetic classifications) may bear a useful relatedness with other biological traits. Based on increasing, but as yet incomplete, evidence it can be suggested that residues of antigenic sites will frequently have some influence on cell receptor recognition (Harber et al., 1995). Amino acid residues of antigenic sites tend to be located at exposed areas on viral capsids and surface proteins, a suitable location to exert influence on the interaction of viruses with receptors. This coincidence of locations may establish some coevolution of antigenicity and host range. In many other cases, however, biological features dependent on one or more nucleotide or amino acid substitutions, will not correspond to any phylogenetic classification based on entire genomes or selected genomic regions.

5. Conclusions

Control of diseases caused by highly variable RNA viruses is not currently successful. For important diseases (such as acquired immune deficiency syndrome (AIDS) or hepatitis associated with hepatitis C virus (HCV)) no vaccines are available. Selection of inhibitor-resistant viral mutants is the norm rather than the exception (Domingo, 1989, Domingo and Holland, 1992, Morse, 1993). Failures have triggered strong claims that a radical change in strategy to combat viral disease is needed (Lederberg et al., 1992, Kunin, 1993, Morse, 1994, Casadevall, 1996, Domingo et al., 1997). Part of the problem lies in the quasispecies nature of the RNA viral agents. Another problem is the still rudimentary knowledge of the immune system and immune responses. Quasispecies offer alternatives for improving disease prevention and treatments (Novella et al., 1995, Domingo et al., 1997). Quasispecies structure is an important problem for viral disease control and should not be minimised on account of the indeterminations found when trying to define compositions of mutant spectra. The evolution of an RNA virus may be rapid or slow, but mutant swarms are always hidden in their replicating genome populations. Research to better understand quasispecies composition and dynamics is urgently needed, since quasispecies are an important step in the natural history of RNA viruses.