Natural non-homologous recombination led to the emergence of a duplicated V3-NS5A region in HCV-1b strains associated with hepatocellular carcinoma

Abstract

Background: The emergence of new strains in RNA viruses is mainly due to mutations or intra and inter-genotype homologous recombination. Non-homologous recombinations may be deleterious and are rarely detected. In previous studies, we identified HCV-1b strains bearing two tandemly repeated V3 regions in the NS5A gene without ORF disruption. This polymorphism may be associated with an unfavorable course of liver disease and possibly involved in liver carcinogenesis. Here we aimed at characterizing the origin of these mutant strains and identifying the evolutionary mechanism on which the V3 duplication relies.

Methods: Direct sequencing of the entire NS5A and E1 genes was performed on 27 mutant strains. Quasispecies analyses in consecutive samples were also performed by cloning and sequencing the NS5A gene for all mutant and wild strains. We analyzed the mutant and wild-type sequence polymorphisms using Bayesian methods to infer the evolutionary history of and the molecular mechanism leading to the duplication-like event.

Results: Quasispecies were entirely composed of exclusively mutant or wild-type strains respectively. Mutant quasispecies were found to have been present since contamination and had persisted for at least 10 years. This V3 duplication-like event appears to have resulted from non-homologous recombination between HCV-1b wild-type strains around 100 years ago. The association between increased liver disease severity and these HCV-1b mutants may explain their persistence in chronically infected patients.

Conclusions: These results emphasize the possible consequences of non-homologous recombination in the emergence and severity of new viral diseases.

Fig 1. Intra-host demographic dynamics for NS5A-dup (A) and wild (B) strains of HCV1-b derived from Bayesian Skyline Plotting at the NS5A locus.

When temporal sampling was available, time was scaled in years; otherwise time is given in coalescence time.

Fig 2. Posterior distributions of mean substitution rates of NS5a for each quasispecies of HCV1-b.

Quasispecies carrying only one V3 domain (wild type) are represented in red; quasispecies carrying two V3 domains are represented in blue.

Fig 3. Divergce of the NS5A-dup strain from wild-type strains.

Ninety percent posterior probability tree inferred from BEAST analysis under the coalescent constant size model with a random relaxed uncorrelated lognormal clock. NS5A-dup strains are indicated in blue. All strains carrying two V3-domains show common ancestry.

Fig 4. Divergence of NS5A-dup strains (in blue) from wild strains (in black).

Ninety percent posterior probability tree inferred from BEAST analysis under the coalescent constant size model with a random relaxed uncorrelated lognormal clock. Age of nodes is mentioned on time-axis. The divergence between wild and duplicated type sequences is estimated to have occurred at the end of 70’s.

Fig 5. The origin of the first V3 region (R1, in green) in NS5A-dup strains differs from that of the second V3 region (R2, in red), which is more closely related to the wild-type region (in yellow).

Ninety percent posterior probability tree inferred from BEAST analysis under the coalescent constant size model with a random relaxed uncorrelated lognormal clock. Time is on the x-axis and represents the divergence time between clades. Divergence between R1 and R2 is estimated to have occurred around 1920.

Fig 6. The origin of the R1 copy of the V3 region (in blue) differs from that of wild-type V3 regions (in black) at the world scale.

R2 copies (in red) have a common origin with wild V3. Ninety percent posterior probability tree inferred from BEAST analysis under the coalescent constant size model with a random relaxed uncorrelated lognormal clock.