The Quasispecies Nature and Biological Implications of the Hepatitis C Virus

Abstract

Many RNA viruses exist as a cloud of closely related sequence variants called a quasispecies, rather than as a population of identical clones. In this article, we explain the quasispecies nature of RNA viral genomes, and briefly review the principles of quasispecies dynamics and the differences with classical population genetics. We then discuss the current methods for quasispecies analysis and conclude with the biological implications of this phenomenon, focusing on the hepatitis C virus.

General Introduction

Due to an error prone replication mechanism, and the ability to survive mutation, hepatitis C virus (HCV) evolves rapidly and has extensive sequence diversity (Choo et al., 1991; Domingo et al., 1985). Viral isolates are divided into six major groups called genotypes, and within the genotypes there are many subgenotypes. Between genotypes there is 66%–69% sequence identity among isolates at the nucleotide level across the entire genome, 77%–80% between subtypes, and 91%–99% within a given subtype (Bukh et al., 1995; Simmonds, 1995). Within an infected individual, HCV circulates as a set of closely related variants referred to collectively as quasispecies (Domingo et al., 1985; Martell et al., 1992; Steinhauer and Holland, 1987b), rather than as a clonal population (Figure 1).

Figure 1.

HCV sequences have been classified into six major genotypes with approximately 65% sequence identity. Within each genotype, sequences are further classified into subtypes with 78% sequence identity. Different isolates within the same subtype share 91–99% sequence identity. Within a host, HCV circulates as a population of very closely related, but not identical, variants called a quasispecies. Genotypes are denoted by numbers (1–6) and subgenotypes by lower case letters.

The sequence heterogeneity in the quasispecies is a common feature of viruses that have a replication intermediate composed of RNA. One of the first physical demonstrations of the quasispecies nature of RNA viruses was revealed during the sequencing of the RNA phage QB by T1-finger printing (Domingo et al., 1978). After re-cloning the phage from a single plaque, the observed fingerprint of the new isolate deviated from the parental clone. When this process was repeated, several different fingerprint patterns were observed that deviated from the original pattern and from each other, usually by one nucleotide, but occasionally by two or more nucleotides. After serial passages of these clones, the original “wildtype” fingerprint pattern was restored. The investigators concluded that the phage population was heterogeneous, but that there was a defined “average” sequence for the population. Today, the existence of quasispecies has been shown in a number of viruses. The biological implications of this phenomenon have been demonstrated in human immunodeficiency virus-1 (HIV) (Wain-Hobson, 1992), hepatitis C virus (HCV) (Forns et al., 1999), foot-and-mouth disease virus (Domingo et al., 1992), vesicular stomatitis virus (VSV) (Steinhauer et al., 1989), polio (Vignuzzi et al., 2006) and others.

The rapid mutation of such viruses is due to the error prone nature of RNA dependent RNA polymerases that generally lack the proof reading ability of DNA polymerases associated with 3′ -> 5′ exonuclease activity (Duarte et al., 1994; Kohlstaedt et al., 2009; Steinhauer et al., 1992) leading to mutation rates of 1 mutation per 10³ to 10⁵ bases copied per replication cycle (Drake and Holland, 1999). In conjunction with small genomic size, such viruses may incorporate one point mutation with each round of replication (Drake and Holland, 1999). The mutations are frequently well tolerated, allowing mutants to survive. As a result, these organisms can exist in a host as a cloud of closely related sequence variants, differing by as little as one nucleotide from a population average sequence. Collectively these variants are referred to as quasispecies.

Many viral factors, such as the error rate of the polymerase, short replication cycle, and compact genome contribute to the generation of the cloud of variants. Additionally, host factors and immune responses exert a selection pressure that contributes to evolution and diversification of the quasispecies. The variation between individual viral genomes in vivo is thought to contribute to persistence, resistance to treatment, tissue tropism, and the failure of experimental vaccines (Holland et al., 1992).

Quasispecies Theory

To explain the adaptability of self replicating RNA elements, Eigen and Schuster first developed mathematical models of quasispecies behavior, based on principles of evolution and information theory (Eigen and Schuster, 1979). The equations describing quasispecies dynamics are out of the scope of this review, but the interested reader is referred to (Bull et al., 2005; Eigen, 1996; Eigen, 2000; Eigen and Schuster, 1979; Wilke, 2005). Quasispecies theory of RNA viruses differs from classical population genetics in a number of its fundamental assumptions about the nature of genomic replication. These include a high mutation rate of the genome, a small genome size, an extremely large (approaching infinite) population of organisms, and an equilibrium state where effects of mutations equal the effects of selection (Comas et al., 2005). According to quasispecies theory, under such conditions in a stable environment, one can ignore the effects of neutral mutations (mutations that neither hurt nor promote the fitness of a variant) and genetic drift that are assumed in classical population genetics (Eigen, 1987; Holmes and Moya, 2002; Jenkins et al., 2001).

In quasispecies theory, the frequency of a particular variant is dependent not only on its inherent fitness, but also on its relatedness to other variants. The closer a particular sequence is to other variants, the more likely that sequence will arise again independently as a result of mutations introduced into other variants by error prone replication. It follows that in a quasispecies, evolution and selection act on the population as a whole, rather than on individuals (Biebricher and Eigen, 2005; Eigen, 1987; Eigen, 1993; Eigen, 1996). Each variant contributes to the fitness of the population. Even a particularly fit variant, will be unable to maintain itself indefinitely in the population due to the high mutation rate leading to reversions of fitness. Studies of VSV in cell culture have demonstrated physically that having a population of closely related variants can allow an individual with low fitness to out-compete an individual of higher fitness, provided the high fitness species is a minor component of the population (de la Torre and Holland, 1990). This has been described as the triumph of the “mean” (average) over the fittest (Gomez and Cacho, 2001). Usually, there exists a dominant sequence that is most frequently represented in the population (Eigen, 1987; Eigen, 1993; Eigen, 1996; Nowak and May, 1992) although this may amount to as little as 10% of the population. Classically, the dominant sequence is assumed to be the one best adapted to the environment in which it was observed.

The applicability of quasispecies theory to the behavior of RNA viruses remains somewhat controversial among evolutionary biologists (Comas et al., 2005; Holmes, 2003; Holmes and Moya, 2002; Jenkins et al., 2001; Moya et al., 2000; Wilke, 2005). Specifically, some argue that in practice the main implication of quasispecies theory, that selection acts on the population rather than on the individual, is invalid because the effects of genetic drift can not be ignored. Furthermore, while sequencing studies have found that over half of the positions in a typical viral genome can be mutated and still lead to a viable offspring (Duarte et al., 1994), in practice, few nucleotide sites in the viral genome are truly neutral and many single mutations confer a lethal, non replicating phenotype. Synonymous sites in the genome have generally been the source of neutrally evolving positions. However, since selection on these sites might arise for reasons other than protein coding including codon bias of the host and restrictions imposed by RNA secondary structures, the effects of supposedly neutral mutations can not be ignored. Additionally, others argue that the mutation rate in RNA viruses is not sufficiently high to establish a population that can be accurately described by quasispecies theory. Studies using systems of “digital organisms,” computer models representing linear RNA viral sequences, suggest that the necessary mutation rates for stable quasispecies development are not attainable in nature (Comas et al., 2005).

Regardless of whether the kinetics of quasispecies theory applies to RNA viruses, their existence in a host as a cloud of related but not identical of viral variants has important biological implications.

Methods of quasispecies analysis

When analyzing a quasispecies, it is useful to begin by determining the consensus sequence of all variants in the population. Obtaining the consensus sequence is relatively straightforward. Viral RNA is isolated from a clinical specimen or from an experimental system, amplified by RT-PCR, and the amplicon is sequenced directly. Automated sequencing machinery typically selects the base that is most common in the amplicon at each position. If two or more bases are present with roughly equal frequencies at a particular position, the automated sequencing machinery may not be able to identify the single most common base. Review of the chromatogram accompanying the sequencing results is useful in resolving such discrepancies and in obtaining a rudimentary sense of the variation in the population.

Cloning and nucleotide sequencing

The sequencing of individual variants in the quasispecies is a multi-step and costly endeavor, but it can provide many details about the composition of the quasispecies. This process begins with amplification of the viral genome using primers that bind to highly conserved areas of the genome surrounding the region of interest. Sequence variations in the primer binding site may prevent the amplification of a minority of the population. The PCR product is cloned into a plasmid vector that can be used to transform bacteria. Transformed bacteria are plated at a low density to allow for the selection of individual colonies, which are assumed to be carrying only one plasmid and therefore only one viral variant. By retrieving and sequencing the plasmid DNA from an individual bacterial colony, one can obtain the sequence of a single variant.

In all reports of quasispecies sequencing, only a small fraction of the viral genome is sampled. Studies of translation in HCV often include analysis of the internal ribosome entry site (IRES), while investigations of viral entry and immune evasion typically focus on the envelope and hypervariable region (HVR1). The core, HVR1, and NS5A genes are often selected for analyses that predict IFN treatment outcomes. When analyzing a single gene or region, it is important to consider that downstream and/or upstream mutations may modulate the effects of mutations in the region of interest. Furthermore, the choice of region to study can greatly impact the conclusions, as demonstrated in studies by Farci and colleagues (described below) who reported an association between increased quasispecies complexity in the HVR1 and HCV disease progression, but did not find a similar association with other regions of the genome (Farci et al., 2000).

Opinions vary on the number of clones that need to be sequenced to appropriately investigate the quasispecies. For mutational frequency and entropy analysis, 20–100 clones per sample are recommended (Gretch and Polyak, 1997). It has been estimated that sequencing 99 clones of HVR1 will identify 95% of all variants present at a frequency of at least 3% in the population (McCaughan et al., 2003). Another estimate suggests 20 clones is sufficient to sample 95% of the major variants (those with at least 10% frequency in the population) (Gretch and Polyak, 1997), however most minor variants may be undetected, leaving out useful information (Gretch et al., 1996). It is important to consider the conservation of the gene or genomic region when deciding how many clones to sequence.

The recent development of massively parallel ultra-deep pyrosequencing has allowed investigators to sequence a much larger fraction of the quasispecies. This increase in genomes sampled comes at the expense of sequence coverage with reads of often less than 200 base pairs. Ultra-deep sequencing works by separating individual DNA molecules from the cDNA amplicon and attaching them to specialized beads. Local in-vitro amplification of each DNA molecule is performed in an emulsion, and beads are then loaded into wells on a plate for pyrosequencing. Light from the pyrosequencing reactions are recorded and used to determine the sequence of the input DNA molecules. This method allows rapid sequence determination of a large number of variants by eliminating the need to separate molecules by cloning in bacterial vectors (Eriksson et al., 2008; Margulies et al., 2005). Ultra-deep sequencing has been used to identify extremely minor variants in the quasispecies that may be related to the development of drug resistance in HIV infected patients (Hoffmann et al., 2007; Simen et al., 2009; Wang et al., 2007) and is emerging as a technology for quasispecies analysis of HCV.

Nucleotide sequencing is essential to studies of transmission, phylogenetics, and outbreaks. During HCV transmission events, a small subset of the population in the initial host is transferred to the new host and then subjected to the pressures of the new host’s genetic background and immune system. The quasispecies in the two individuals are related but genetically distinct, as shown by experiments involving infected chimps, vertical transmission, and transplant recipients (Kudo et al., 1997; Murakami et al., 2000; Ni et al., 1997; Power et al., 1995; Prince et al., 2004; Wyatt et al., 1998). Sequencing is also useful in demonstrating replication and tropism of the virus for particular host compartments (Fishman et al., 2008).

Although nucleotide sequencing provides very detailed information about the individual members of the quasispecies, this method is subject to some notable artifacts. Sequence errors may be introduced by the reverse transcriptase and DNA polymerase during amplification of the template. DNA polymerases with proofreading activity should always be used when performing PCR for quasispecies analysis to minimize such errors. Studies of quasispecies complexity can be skewed by template re-sampling during PCR when the input nucleic acid concentration is low. When this occurs, the apparent complexity is lower than the true value. In addition, the major variants of the population may be obscured by the repeated copying of minor variants. Given the error rate of the viral polymerase, and the length of the fragment of interest, one can estimate the number of species to expect based on the initial viral RNA concentration (Airaksinen et al., 2003). Significant deviations from this value should be viewed cautiously. In practice, if a 1:100 dilution of the viral RNA is sufficient to generate an RT-PCR product, then template re-sampling is unlikely to occur (Domingo et al., 2006).

Analyzing quasispecies sequences with bioinformatics

Once the sequences of the variants are obtained, a number of bioinformatics based analyses can be performed. The proportion of synonymous and nonsynonymous substitutions in the sequences is often examined. Synonymous or silent mutations (those that do not change the amino acid sequence of the protein) are traditionally considered evolutionarily neutral. Due to the degeneracy of the genetic code, synonymous substitutions can be made for all amino acids except methionine and tryptophan, which each have only one codon. Nonsynonymous substitutions change the amino acid sequence of the protein. The ratio of nonsynonymous substitutions per eligible site (d_n) to synonymous substitutions per eligible sites (d_s) is an indicator of the strength of the positive or negative selection acting on the population. A d_n/d_s ratio greater than one is in indicative of “positive selection” acting on the population (Kimura, 1977). Positive selection results in the increased frequency of a nonsynonymous mutation in the population due to the reproductive advantage it confers. Conversely, a ratio of less than one indicates “negative selection,” specifically that nonsynonymous mutations are unfit and have been culled.

Although d_n/d_s ratios are frequently reported for quasispecies populations, the validity of these ratios as an indicator selective pressure is controversial in these groups. The use of d_n/d_s ratios as an indicator of selective pressures rests on the assumption that nonsynonymous changes are fixed in the sequence. However, when comparing closely related sequences of the same rapidly evolving species, the majority of observed nonsynonymous substitutions are transient (Rocha et al., 2006). Studies have shown that the relationship between d_n/d_s ratios and the type of selective pressure acting on the population is dependent on the divergence of sequences within the sample. Sequence samples drawn from different lineages will have d_n/d_s ratios that accurately reflect the nature of the selective pressure acting on the population. However, d_n/d_s ratios calculated from sequence samples drawn from a single population (for example, a quasispecies) are not an indicator of the type of selection (Kryazhimskiy and Plotkin, 2008). Since quasispecies are inherently dynamic, selection pressures as reflected by d_n/d_s ratios can be skewed by a delay in the removal of variants with mutations that reduce viral fitness. In addition, sites in the genome may be subjected to pressures acting on RNA structures embedded in coding regions and/or on overlapping coding sequences that lead to altered d_n/d_s ratios (Domingo, 2003; Xing and Lee, 2006).

Other parameters commonly of interest when studying the quasispecies composition are the complexity and diversity of the population. Complexity refers to the number of different sequences present in the population. Usually complexity is presented as either the ratio of unique variants to the total number of clones analyzed or as the Shannon entropy:

$$-\frac{{\displaystyle \sum _{i=1}^n}{f}_i(ln{f}_i)}{N}$$

where n is the number of different species identified, f_i is the observed frequency of the particular variant in the quasispecies, and N is the total number of clones sequenced. Dividing by N normalizes for differences in the number of clones analyzed. Statistical comparisons of complexity between two groups are usually made using the Wilcoxon rank sum test or other non parametric tests of the median. A broad spectrum of variants (high complexity) may indicate similar fitness among viral variants and a long duration of stability of the population. Similarly, low complexity may reflect larger differences in fitness and a more recent change in the environment. Alternatively, low complexity may reflect a lack of selection pressures.

The diversity of a quasispecies refers to the relatedness of individuals within the population. The most basic measure of diversity is the average hamming distance, that is, the number of mutated positions in a particular sequence relative to a dominant sequence, consensus sequence, or other reference. More often, diversity is measured using a genetic distance algorithm such as the Jukes-Cantor or Kimura-2-parameter metric, that consider additional factors such as the probabilities of transitions (purine to purine or pyrimidine to pyrimidine substitutions), transversions (pyrimidine to purine and vice versa substitutions), and the global nucleotide or amino acid composition of the sequences. Genetic distances within and between quasispecies are often used to generate multiple sequence alignments and phylogenetic trees. Sequences that cluster or are assigned to the same clades are most closely related to each other. The strength of the clustering of sequences on a phylogenetic tree can be evaluated by repeatedly randomizing the sequences and recreating the tree (bootstrapping).

Transmission and outbreak studies often measure genetic diversity and create phylogenetic trees of the quasispecies of different patients. Usually patients are infected with a single quasispecies that is recognizable as a unique clade on a phylogenetic tree when compared with sequences from other individuals. Infection of multiple people with a single isolate typically leads to divergence in the new hosts. Multiple, closely-related quasispecies evolve from the original quasispecies (Ray et al., 2005). As a result of super-infection, genetically distinct quasispecies can circulate in an individual patient.

Many studies relate intra-patient quasispecies diversity to other clinical features of infection such as progression of disease and response to treatment. Since there is no one standard way to measure diversity, it is difficult to compare results between these studies. In addition, the observed diversity of the quasispecies is partially dependent on the region of the genome analyzed because the conservation of viral genomes is non-uniform.

Single stranded conformation polymorphism (SSCP)

While nucleotide sequencing provides valuable information on the composition of the individual variants, a popular, rapid alternative is SSCP. Denatured single stranded PCR products are separated on a non-denaturing gel at constant temperature by electrophoresis to resolve different variants based on the secondary structure of the nucleic acid (Hayashi, 1991; Hongyo et al., 1993; Orita et al., 1989). It was originally though that variants differing by as little as one base may adopt different secondary structures, and therefore have measurable differences in gel mobility (Melcher, 2000), although more recent studies have found that single nucleotide changes may not resolve as different bands (Vera-Otarola et al., 2009). The gel bands are visualized either by a radiolabel on the control sequence, Southern hybridization, or staining. The number of bands in the lane reflects the number of variants in the sample. This method can detect only those variants that comprise approximately 3% or more of the population (Laskus et al., 1996). Extremely minor variants are unlikely to be detected.

Many variables can affect the performance of SSCP (reviewed in (Hayashi and Yandell, 1993)). For example, the mobility of DNA is affected by the temperature and consistency of the polyacrylamide gel. Reproducible results require electrophoresis to be performed at the same temperature every time. Long fragments may not resolve as clearly as shorter fragments, the optimal range being between 150 and 300 base pairs, although the addition of glycerol can add resolution when larger fragments are used (Kukita et al., 1997). Other factors affecting the sensitivity of SSCP include the concentration of the DNA fragments, the G/C content of the nucleic acids, and properties of the buffer solution.

While SSCP can be a convenient, inexpensive tool for assessing the general composition of the quasispecies, there are important drawbacks. Information on the complexity of the population is obtained by inspection of the gel, but more detailed knowledge of the diversity and specific information on the nature of mutations present in the quasispecies requires the excision and sequencing of the individual bands on the gel, which is both labor intensive and expensive. In addition, because the isolation of individual variants by SSCP is sensitive to a variety of parameters, experimental conditions must be optimized empirically.

Heteroduplex gel shift assays

Heteroduplex gel shift assays are similar in concept to SSCP. This method has been applied to HIV (Delwart et al., 2003) and HCV quasispecies (Wilson et al., 1995). The advantages and drawbacks of heteroduplex gel shift assays are similar to SSCP methods in that the entire population can be readily sampled, but there is no information about the specific substitutions in the population.

There are two types of heteroduplex shift assays: The homo duplex tracking assay (HTA) and clonal frequency analysis (CFA). In both cases, the viral RNA is first isolated and amplified, and a radio-labeled probe is synthesized to hybridize to the region of interest. In the HTA assay, the labeled probe is hybridized to the mixed population of variants present in the PCR amplicon. The heteroduplex products are resolved by electrophoresis under partially denaturing conditions. The different variants in the heterogeneous amplicon will have varied migration due to mismatches with the probe resulting in bubbles that hinder gel mobility and create distinct bands. The number of distinct bands in the lane reflects the number of variants in the quasispecies. In the CFA assay, the variants are separated by cloning into plasmids (as is done for quasispecies sequencing). A number of clones are individually hybridized to the probe. The resulting duplexes are resolved in neighboring lanes using gel electrophoresis. Genetic distance can be calculated by measuring the heteroduplex mobility ratio (HMR): the distance in millimeters from the origin of the gel to a specific heteroduplex divided by the distance in millimeters from the origin to the homoduplex control (the band with the farthest migration). The HMR is calculated for each variant and then averaged to give a measure of genetic diversity for the population. The reduced migration of a mismatched heteroduplex relative to a homoduplex, and the resulting HMR value, is proportional to the number of different nucleotides between two variants (Delwart et al., 1993; Polyak et al., 1997; Wilson et al., 1995). HMR values for two populations can be compared statistically by t-tests (Polyak et al., 1998). The ability to estimate the population diversity is an advantage of heteroduplex gel shift assays over SSCP. However, these assays require variants of a uniform length to differ in sequence by about 1.5% for accurate resolution (Delwart and Gordon, 1997).

Matrix-assisted laser desorption ionization-time of flight mass spectroscopy (MALDI-TOF)

In this assay, the extracted RNA templates are reverse transcribed and amplified by RT-PCR, and subsequently used to synthesize tagged peptides. The peptides are subjected to MALDI-TOF mass spectrometry to identify the amino acid composition of the peptides. This method of analysis is inexpensive and rapid relative to sequencing multiple cloned fragments (Farci et al., 2000; Farci et al., 2002; Yea et al., 2007). An advantage of the MALDI-TOF method of quasispecies analysis over other rapid methods such as SSCP and heteroduplex gel shift assays is the knowledge of the amino acid composition of the fragments, from which information can be deduced about nonsynonymous nucleotide mutations. However, variants comprising less than 5% of the population are not identified.

Biological implications of a quasispecies distribution for RNA viruses

Viruses that circulate as a quasispecies present a unique set of challenges to the host. Variants may differ in their biological properties such as virulence, ability to escape the immune system, resistance to antiviral therapies, and tissue tropism. A specific variant within the quasispecies can have a phenotype that differs from the majority of the population. If such variants arise de novo they can change the course of disease (Duarte et al., 1994). For example, in measles virus infection, certain variants contribute to high virulence, central nervous system tropism, and distinct disease manifestations (Steinhauer and Holland, 1987a). In lymphocytic choriomeningitis virus, a single amino acid change can alter the tissue tropism (Dockter et al., 1996). The quasispecies nature of viruses such as HCV and HIV may contribute to the challenge of vaccine development because the use of live attenuated viruses in vaccines is risky due to the potential of these viruses to mutate rapidly and become virulent.

The major implications of a quasispecies distribution are discussed below with examples from studies of HCV.

Transmission and quasispecies divergence

For viruses that circulate as a quasispecies, there are usually multiple variants present at the time of transmission of infection. Transmission of all variants within a quasispecies is not uniform (Gao et al., 2002; Sugitani and Shikata, 1998; Weiner et al., 1993). Often transmission results in a population bottleneck, with only a small fraction of the variants in the original quasispecies passing to the new host. The dominant variants in the inoculum are often poorly adapted to the new environment. As a result, a minor variant in the quasispecies often becomes dominant in the new host. When multiple variants successfully make the transition into the new host, there is high quasispecies diversity during the early phase of infection (Herring et al., 2005).

Within an infected individual HCV quasispecies diversity can vary greatly. Herring et al., reported sequences differing by 1 to 7.8% at the nucleotide level in HVR1 in the quasispecies of 12 subjects during the early phase of infection. A study of an outbreak in which patients were infected from a single source showed that variants present in the inoculum may give rise to divergent populations within an individual patient. The two populations can be as different from each other as they are from the progeny sequences in other patients (Ray et al., 2005). However, the generation of two distinct populations from a single infection event within an individual patient is uncommon.

Compartmentalization

Analysws of variants isolated from different body compartments show that the members of the quasispecies are not randomly distributed. Variants with different tissue tropisms and compartmentalization of genomes have been observed for a number of RNA viruses including HIV and HCV. Sequence variants that are restricted to a particular body compartment have been found in the serum (Cabot et al., 2000; Jang et al., 1999; Shimizu et al., 1997), peripheral blood mononuclear cells (PBMCs) (Ducoulombier et al., 2004; Laskus et al., 1998; Navas et al., 1998; Roque-Afonso et al., 2005; Shimizu et al., 1997), CNS (Fishman et al., 2008; Forton et al., 2004; Radkowski et al., 2002; Vargas et al., 2002), and other extrahepatic sites in specimens from patients with HCV, suggesting that some portions of the quasispecies may replicate in isolation from other portions. Individuals infected with multiple distinct quasispecies may show complete segregation of the two populations. In extreme cases, sequences of one HCV genotype are isolated from a specific compartment (e.g., liver) and sequences of a different genotype are isolated from a different compartment (e.g., PBMCs). We previously described a patient infected with both genotype 1a and genotype 1b HCV. Genotype 1b variants were found exclusively in the liver, while genotype 1a variants were found in the liver, plasma, and brain tissue (Fishman et al., 2008).

Tissue-specific mutations often occur in the HCV IRES, presumably because the cellular proteins that facilitate IRES function differ between cell types, driving the evolution of variants adapted for the local set of proteins. Investigations have shown that IRES mutations alter translation efficiency (Forton et al., 2004; Gallegos-Orozco et al., 2006). It has been suggested that inefficient variants evolve as a means of down-regulating viral protein expression, thereby evading the immune system (Gallegos-Orozco et al., 2006; Laporte et al., 2000; Laporte et al., 2003); however, it is also possible that IRES mutations promote efficient translation in a specific cell type that was not included in the assay.

The demonstration of sequence compartmentalization of a single quasispecies is usually shown by construction of a phylogenetic tree where variants from a given compartment cluster together, although complete segregation of variants by compartment in separate clades is rare. It is also possible to demonstrate that the genetic distances observed between variants isolated from different compartments are significantly greater than the genetic distances observed between variants in the same compartment using Mantel’s test (see methods of (Roque Afonso et al., 1999)). This method has been applied to compartmentalization studies in both HCV (Roque Afonso et al., 1999) and HIV (Poss et al., 1998).

Viral persistence and progression of disease

A few RNA viruses, including HCV, can establish a chronic infection. It is thought that changes in antigen epitopes may contribute to persistence by allowing the virus to escape from the adaptive immune system. According to this hypothesis, in the absence of immune pressure, there should be little or no antigenic diversity in the viral population. Indeed, one small study of HCV patients with agammaglobulinemia, found limited quasispecies diversity (Kumar et al., 1994). More recent studies of HCV patients undergoing liver transplantation for HCV-related cirrhosis showed that post transplant viral complexity and diversity was lower in HVR1 than pre-transplant, suggesting that immunosuppression can decrease the introduction of new variants into the quasispecies (Feliu et al., 2004; Moreno et al., 2003; Schvoerer et al., 2007). Furthermore, in patients co-infected with HIV, immunosuppression is associated with reduced quasispecies diversity (Canobio et al., 2004; Toyoda et al., 1997).

However, the diversification of quasispecies due to immune pressure remains controversial, and may not be true of all viruses that circulate as a quasispecies. Studies with foot-and-mouth-disease virus and influenza virus show that antigenic diversity and fluctuation can occur even in the absence of immune pressure (Borrego et al., 1993; Domingo et al., 1993; Umetsu et al., 1992). It is likely that the presence of variants whose proteins are not recognized by the immune system contributes to persistence of the virus.

It remains unclear if the increasing complexity and diversity of the quasispecies, and in particular immune escape mutants are the cause or result of chronic infection. Studies of HCV show that progression from acute to chronic infection is associated with an increase in complexity in HVR1 sequences of the quasispecies (Farci et al., 2000). In patients with transfusion acquired HCV, viral clearance was associated with stasis of the quasispecies during acute phase, while progression to chronic infection was associated with evolution of the quasispecies (Farci et al., 2000). In serum samples taken before seroconversion, there was a significant decrease in the diversity of the quasispecies relative to pre-seroconversion serum samples in patients who cleared the virus spontaneously. No such decrease in quasispecies diversity was observed in those who progressed to develop chronic infection. These changes in the quasispecies were observed only in the HVR1 region, underscoring the importance of HVR1 in HCV-host interactions. This study also illustrates how the choice of the region under investigation can impact the results of quasispecies analysis. Furthermore, GB virus B, a distant relative of HCV, has limited quasispecies diversity during acute infection, and rarely progresses to chronic infection and persistence (McGarvey et al., 2008).

During chronic HCV infection, many factors affect the rate of liver disease progression. Advanced disease is associated with a longer duration of infection, male sex, older age at time of infection, HIV and/or hepatitis B virus co-infection, and high consumption of alcohol (Domingo and Gomez, 2007). Interestingly, the disease progression rate is independent of viral load in the circulation and viral genotype. Studies of the relationship between the diversity/complexity of the quasispecies and liver damage, progression to cirrhosis and hepatocellular carcinoma, and extrahepatic manifestations have yielded conflicting results (Arenas et al., 2004; Canobio et al., 2004; Gonzalez-Peralta et al., 1996; Hayashi et al., 1997; Honda et al., 1994; Koizumi et al., 1995; Lopez-Labrador et al., 1999; Naito et al., 1995; Qin et al., 2005; Sookoian et al., 2000; Vallet et al., 2007; Yuki et al., 1997). High levels of alanine aminotransferase (ALT) are indicative of liver injury. Rothman et al. reported an association between increased complexity of the HCV quasispecies in HVR1 and ALT levels (Rothman et al., 2005). In a study of children, Farci and colleagues found no association between the diversity and complexity of HVR1 sequences and ALT levels before seroconversion, but found that high levels of ALT after conversion correlated with a decrease in complexity and diversity of the quasispecies relative to the pre-seroconversion measurements (Farci et al., 2006). Over time, children with low ALT values had changing populations of virus, with sequential shifts in the populations, while children with high ALT values did not have temporally segregated populations of virus.

Drug and IFN response

The standard therapy for HCV infection consists of weekly injections of pegylated IFN alpha and daily doses of ribavirin for 6 months or a year. The duration depends on the genotype. This regimen leads to a sustained virological response in approximately 42% of patients infected with genotype 1 and 82% of patients infected with genotypes 2 and 3 (Fried et al., 2002; Hadziyannis and Koskinas, 2004; Manns et al., 2001; Pawlotsky, 2003). Many studies have sought to predict the outcome of IFN treatment based on pre-treatment sequence features including specific viral mutations and quasispecies parameters (diversity and complexity). The presence of a quasispecies may contribute to the low success rate of treatment since individual variants may have different levels of natural IFN induction (Pellerin et al., 2004).

Young patients with little or no liver damage, low viral load, non-genotype 1, and a short duration of infection are most likely to achieve a sustained virological response with pegylated IFN alpha/ribavirin treatment (Domingo and Gomez, 2007). Many recent studies associate pre-treatment high complexity and diversity of the HCV quasispecies in HVR1 in chronic infection (Chambers et al., 2005; Kumar et al., 2008; Moreau et al., 2008; Puig-Basagoiti et al., 2001; Sarrazin et al., 2000; Thelu et al., 2001; Ueda et al., 2004) with treatment failure. A decrease in quasispecies complexity over the initial treatment period is associated with sustained virological response (Farci et al., 2002; Salmeron et al., 2008), while conversely no change in the composition of the population is associated with treatment failure (Pawlotsky et al., 1998b; Pawlotsky et al., 1999; Quesnel-Vallieres et al., 2008). Interestingly, some analyses of quasispecies in the NS5A region associate high complexity and diversity with interferon sensitivity (Enomoto et al., 1995; Pawlotsky et al., 1998a; Puig-Basagoiti et al., 2001; Puig-Basagoiti et al., 2005; Saiz et al., 1998; Sarrazin et al., 2002; Ueda et al., 2004; Veillon et al., 2007), however this finding has not been universal (Puig-Basagoiti et al., 2005).

Induction of error catastrophe as an antiviral strategy

In viruses that circulate as a quasispecies, the polymerase error rate is such that on average one point mutation is made per replication cycle (Drake and Holland, 1999). This mutation rate works to the advantage of the virus. Among mammalian viruses, the error rate contributes to fitness by helping the virus evade the immune system (Kamp and Bornholdt, 2002). The beneficial effects of the viral error rate have been demonstrated in a study of poliovirus (Vignuzzi et al., 2006). When the fidelity of the polio polymerase was increased, leading to the generation of fewer variants, the virus showed reduced pathogenicity and neurotropism. The addition of mutagens to the culture media restored fitness (Vignuzzi et al., 2006).

Although RNA viruses incorporate new mutations at extremely high rates, there is a limit to their capacity to absorb mutations. When this capacity is exceeded, the integrity of the genome is compromised, and the virus is eliminated through a process known as error catastrophe. The fact that viral RNA replicases operate at an error level near the threshold for error catastrophe suggests that pharmaceuticals that increase the error rate would be potentially useful anti-viral agents. Holland et al., first applied this concept to the poliovirus by demonstrating a negative effect on replication with the addition of mutagens (Holland et al., 1990). Mutagens have since been used to treat infection with HCV, HIV, and many other RNA viruses (reviewed in (Anderson et al., 2004; Domingo et al., 2005)). A number of hurdles remain in the development of antiviral therapies based on enhanced mutagenesis. The compounds are often toxic and impact cellular enzymes as well as viral polymerases causing mutation of the host genome. In addition, the concentration of the mutagenic agent must be maintained at the optimal level, as suboptimal levels would lead to the selection of resistance mutations or the appearance of variants with new biological properties (Domingo, 2003).

Implications to New Antiviral Strategy

Unfortunately, the standard of care therapy for HCV is expensive, arduous, fraught with side effects, and ineffective in about 50% of patients with genotype 1 HCV. Recent efforts to improve treatments for HCV have centered on the development of therapies to inhibit specific steps in the HCV life cycle. Over 30 agents that inhibit viral entry, proteolytic cleavage, RNA replication, and assembly are in the pipeline (reviewed in (Thompson and McHutchison, 2009)). However, due to the quasispecies nature of HCV, it will be difficult to develop durable small molecule inhibitors of HCV. Variants harboring resistant mutations to protease inhibitors have been observed in treatment naïve subjects (Bartels et al., 2008; Sarrazin et al., 2007). The emergence of minor variants with drug resistant phenotypes can be anticipated and has been observed in early trials of the STAT-C (specifically targeted anti-viral therapy for HCV) drugs. Although resistance mutations have generally been associated with reduced viral fitness relative to the wild-type (He et al., 2008), these variants can become major species during treatment. As a consequence, resistant mutants are likely to become more prevalent in the population, reducing the success rate of newly developed drugs. Combination therapy of protease inhibitors with interferon and ribavirin should increase the likelihood of achieving a sustained virological response.

Conclusions

The quasispecies nature of RNA viruses, and HCV in particular, has significant implications for the behaviors of these viruses in vivo. The exact mechanisms by which the variants in the population contribute to compartmentalization, viral persistence and progression, and transmission events remain open questions.

Abbreviations

HCV

hepatitis C virus

IFN

interferon

IRES

internal ribosome entry site

HVR1

first hypervariable region

SSCP

single stranded conformation polymorphism

HIV

human immunodeficiency virus-1

HTA

homo duplex tracking assay

CFA

clonal frequency analysis

HMR

heteroduplex mobility ratio

MALDI-TOF

matrix-assisted laser desorption ionization-time of flight

PBMCs

peripheral blood mononuclear cells

ALT

alanine aminotransferase