Quantitative Estimation of Viral Fitness Using Pyrosequencing™

The detection of sequence changes in a virus population is generally accomplished by the subcloning of individual genomes and the subsequent sequence analysis of a statistically significant number of isolated samples. This method is useful for the identification of mutant alleles arising from drug resistance or reversion from a defective phenotype but is not practical as a protocol to track sequence variation quantitatively over time or to compare the selective advantage of a mutation carefully. The recent development of Pyrosequencing (Pyrosequencing AB, Uppsala, Sweden) (1,7) now allows the evaluation of allele frequency with significant ease and accuracy (6,9) to make quantitative assessment of mutant fitness probable. Recently, O’Meara et al. (8) described the use of Pyrosequencing to detect resistance mutations to human immunodeficiency virus (HIV) protease inhibitors from clinical samples of viral RNA and proviral DNA. In a study of four HIV-infected patients undergoing multidrug regimens, viral RNA samples were obtained and probed for the presence of previously known single nucleotide changes (at 33 different amino acid codons) that confer drug resistance. As the estimation of allele frequency was only semi-quantitative (i.e., 0%, 25%, 50%, 75%, and 100%), many potential sequences may not have been detected or monitored. Although this protocol worked well for the basic detection of specific mutant sequences emerging in the population, the experiments could not quantitatively estimate the rate of change of the mutant allele in a complex population. An accurate measurement of this rate would allow for a ranking of replication fitness of one mutant allele over another, leading to a deeper understanding of the emergence of one population over another (3). This report describes the tracking over time of a mutant allele in a mixed population in a competitive fitness study using Pyrosequencing technologies. To develop this approach, we used a model system, the hepatitis C virus (HCV) replicon system (5), to evaluate the dynamics of RNA genomes by following SNPs. A Huh-7 cell line was created containing a G418-selectable RNA replicon containing the termini and nonstructural genes of the HCV genome, in a manner previously described (2,5). During the establishment of these HCV replicon cell lines, Sanger sequencing of RT-PCR products amplified from total RNA identified several SNPs. To measure the fitness of these SNPs on replication of these genomes, a cell line bearing a single nucleotide variation of the parental sequence was mixed with cells containing the original, “wild-type” sequence. The change in the population over time was monitored by the quantitative analysis of this position, a C in the mutant sequence and a T in the wild-type. The number of replicon genomes/cell for each starter cell line (wild-type, 3100 genomes/cell; mutant, 2300) was measured by quantitative real-time RT-PCR targeting the HCV 5′ nontranslated region (4). Equal numbers of cells (5 × 104 from each culture) were combined initially and then Benchmarks