Rapid sequence change and geographical spread of human parvovirus B19: comparison of B19 virus evolution in acute and persistent infections

Abstract

Parvovirus B19 is a common human pathogen maintained by horizontal transmission between acutely infected individuals. However, B19 virus can also be detected in tissues throughout the life of the host, although little is understood about the nature of such persistence. In the current study, we created large VP1/2 sequence data sets of plasma- and tissue (autopsy)-derived variants of B19 virus with known sample dates to compare the rates of sequence change in exogenous virus populations with those in persistently infected individuals. By using linear regression and likelihood-based methods (such as the BEAST program), we found that plasma-derived B19 virus showed a substitution rate of 4 x 10(-4) and an unconstrained (synonymous)-substitution rate of 18 x 10(-4) per site per year, several times higher than previously estimated and within the range of values for mammalian RNA viruses. The underlying high mutation frequency implied by these substitution rates may enable rapid adaptive changes that are more commonly ascribed to RNA virus populations. These revised estimates predict that the last common ancestor for currently circulating genotype 1 variants of B19 virus existed around 1956 to 1959, fitting well with previous analyses of the B19 virus "bioportfolio" that support a complete cessation of genotype 2 infections and their replacement by genotype 1 infections in the 1960s. In contrast, the evolution of B19 virus amplified from tissue samples was best modeled by using estimated dates of primary infection rather than sample dates, consistent with slow or absent sequence change during persistence. Determining what epidemiological or biological factors led to such a complete and geographically extensive population replacement over this short period is central to further understanding the nature of parvovirus evolution.

FIG. 1.

Variability of ORF1 and ORF2 coding sequences measured as mean pairwise synonymous distances between sequences of the same genotype (A) and different genotypes (B) (note the different y axis scales between panels A and B). Analysis was carried out using 201-base fragments increasing by 24-base increments across each coding region. ARF, alternate reading frame.

FIG. 2.

Phylogenetic analysis of plasma-derived (A) and autopsy/biopsy sample-derived (B) B19 virus sequences from this study (circles) and published sources (diamonds). Trees were constructed from pairwise J-C-corrected nucleotide distances by neighbor-joining. Symbols were colored based on the calendar year of collection (see the key). The robustness of groupings was calculated by bootstrap resampling of 1,000 replicates of the data; values of ≥70% are shown.

FIG. 3.

Relationship between sample date and divergence (MCL-corrected pairwise distances) between the sequence of the reference strain of B19 virus, PVBPRO (collected in 1973), and plasma-derived (A) and autopsy/biopsy-derived (B) sequences. Previously published (open circles) and newly obtained (filled circles) sequences are shown. Each graph includes a line of best fit, calculated according to the linear region.