Recombination in viruses: mechanisms, methods of study, and evolutionary consequences

Abstract

Recombination is a pervasive process generating diversity in most viruses. It joins variants that arise independently within the same molecule, creating new opportunities for viruses to overcome selective pressures and to adapt to new environments and hosts. Consequently, the analysis of viral recombination attracts the interest of clinicians, epidemiologists, molecular biologists and evolutionary biologists. In this review we present an overview of three major areas related to viral recombination: (i) the molecular mechanisms that underlie recombination in model viruses, including DNA-viruses (Herpesvirus) and RNA-viruses (Human Influenza Virus and Human Immunodeficiency Virus), (ii) the analytical procedures to detect recombination in viral sequences and to determine the recombination breakpoints, along with the conceptual and methodological tools currently used and a brief overview of the impact of new sequencing technologies on the detection of recombination, and (iii) the major areas in the evolutionary analysis of viral populations on which recombination has an impact. These include the evaluation of selective pressures acting on viral populations, the application of evolutionary reconstructions in the characterization of centralized genes for vaccine design, and the evaluation of linkage disequilibrium and population structure.

Keywords: Linkage disequilibrium; Mutation rate; Population structure; Reassortment; Recombination; Recombination rate.

Fig. 1

Replication and recombination steps in herpes simplex virus. Both processes are intimately linked. Although the re-circularization of linear DNA has not demonstrated experimentally, several lines of evidence suggest that it must occur. Three ori sites are the distributed along the DNA genome (2 oriS and 1 oriL), but in this figure we only show the interaction of the replication complex with one site. In the oriS (45 bp) there are two fragments with high affinity to UL9 (box1 and box2) characterized by the sequence TTCGCAC, whereas in oriL (144 bp) two box1 are present. It has been suggested that these origins may function differently during lytic and latent infection.

Fig. 2

Pipeline for the estimation of local dN/dS accounting for recombination. (1) Given a coding DNA alignment, recombination breakpoints can be detected, for example with GARD (HyPhy, Datamonkey web server) or RDP3. Partition alignments can be produced by splitting the original alignment according to the recombination breakpoints. (2) A substitution model can be selected for each partition alignment using for example jModelTest (Posada 2008) or HyPhy. (3) Given the substitution model, a phylogenetic tree can be reconstructed for each partition alignment. (4) Finally, the dN/dS estimation at each codon site can be performed by using the corresponding partition alignment and its substitution model and phylogenetic tree. This whole procedure has been implemented in the Datamonkey web server.

Fig. 3

The extent of LD in real populations is expected to decrease with both time and recombination rate between markers, so that the larger the recombination rate the faster the decay in LD. Blue line: low recombination (r = 0.03), Pink line: high recombination (r = 0.3). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

In the case where some recombinant forms (Ab) present lower fitness than others (aB), the outcome of secondary contact between both viral populations will vary from the stable co-existence of two strains within intermediate populations (no recombination: r = 0) to the replacement of the original strains by a new recombinant virus (r = 0.5).