Poxviruses deploy genomic accordions to adapt rapidly against host antiviral defenses

Abstract

In contrast to RNA viruses, double-stranded DNA viruses have low mutation rates yet must still adapt rapidly in response to changing host defenses. To determine mechanisms of adaptation, we subjected the model poxvirus vaccinia to serial propagation in human cells, where its antihost factor K3L is maladapted against the antiviral protein kinase R (PKR). Viruses rapidly acquired higher fitness via recurrent K3L gene amplifications, incurring up to 7%-10% increases in genome size. These transient gene expansions were necessary and sufficient to counteract human PKR and facilitated the gain of an adaptive amino acid substitution in K3L that also defeats PKR. Subsequent reductions in gene amplifications offset the costs associated with larger genome size while retaining adaptive substitutions. Our discovery of viral "gene-accordions" explains how poxviruses can rapidly adapt to defeat different host defenses despite low mutation rates and reveals how classical Red Queen conflicts can progress through unrecognized intermediates.

Figure 1. Experimental evolution of vaccinia virus

(A) Vaccinia encodes two host-range factors, E3L and K3L that inhibit antiviral Protein Kinase R (PKR). (B) Vaccinia was repeatedly passaged through HeLa cells in triplicate (see methods). After each infection, virus titer was measured, an aliquot was reserved for the “fossil record”, and fresh cells were infected. Red dots represent viruses in which a theoretical adaptation arises in the population and increases in frequency. (C) With passaging of ΔE3L virus, replicate populations A-C gained the ability to replicate in HeLa cells as judged by viral titers assayed over the course of the experiment. Dotted horizontal lines show the average titer of parental virus (ΔE3L) and the wildtype Copenhagen strain of vaccinia (WT). Because E3L was replaced by lacZ we could also use β-gal activity assays as a proxy for measuring virus replication (inset). Gains in replication were corroborated by measurements of β-gal activity, shown in arbitrary units, from parental virus and replicates A-C at passage 10 (p10; inset). Data are represented as mean +/− SEM.

Figure 2. Rapid evolution of copy number amplification

(A) Genomic DNA traces of parent and virus replicates A-C at passage 10, showing relative depth of sequence coverage across the entire ~180kb vaccinia genome, except for the inverted terminal repeat regions (ITR; ITRs are shown in Figure S4). Increased K3L sequence coverage was observed in replicates A-C (top right panel; also see Figure S1, S2) while coverage of E3L was missing in all populations (lower right panel). Polymorphisms were detected in the K3L, E9L, and H2R genes as indicated (also see Table S2). We also detected 15 nucleotide differences between the parent and reference Copenhagen strain, which were unrelated to our protocol (Goebel et al., 1990)(Table S6). (B) Southern blots of the K3L locus of viral replicates A, B, and C during the course of experimental evolution from passages 1-10 reveals rapid gene copy number amplification of K3L. Amplification may be a consequence of recombination-driven tandem duplications (Figure S3). (C) Diagrammed is the parental K3L locus and the major duplication breakpoints near K3L that led to copy number amplification (also see Table S3). We also found duplications at the highly repetitive termini of the vaccinia genomes, consistent with previous observations (Moss et al., 1981)(see Figure S4). However, these terminal expansions were also present in the parental strain and therefore not a consequence of our experiments.

Figure 3. Overexpression of K3L protein in adapted strains is necessary and sufficient for increased viral fitness

(A) Western blot analysis of vaccinia and mock infected HeLa cells with wildtype (WT), parental (ΔE3L), and virus replicates A-C at passage 10 for levels of K3L, vaccinia proteins, and actin. (B) siRNA knock-down of K3L in infected HeLa cells. Viral replication was measured by β-gal activity assay of virus and mock infected cells transfected with K3L-specific siRNAs (1-3) or a non-targeting control. Western blot shows levels of K3L, vaccinia proteins, and actin. (C) β-gal activity from HeLa cells infected with various vaccinia viruses. Wildtype vaccinia virus, ΔE3L parental virus, and passage 10 virus from replicate C were compared to plaque-purified clones with either a single copy H47R-K3L (1), a single copy WT-K3L (3), or a multi-copy K3L virus with both WT and the H47R substitutions (2). Numbers correspond to plaque-purified viruses described in Figure 6. Data are represented as mean +/− SEM.

Figure 4. K3L expansions precede the appearance of the H47R substitution

PCR with genomic samples from the “fossil record” of replicate C with primers amplifying duplications of K3L is consistent with Southern blot analysis. Sequencing of duplicated K3L regions reveals that the appearance of the mutation leading to the H47R substitution is only appreciable after duplications of K3L appeared.

Figure 5. Low frequency gene duplications in vaccinia genomes

Of the primers tested (Table S3), duplications at several loci near the termini of the genome were amplified and sequenced (Table S4). Breakpoint regions are drawn to scale highlighting the location of each duplication point. One breakpoint observed in K3L leads to a fusion product in place of the wildtype copy. The C20L and B26R regions are mirrored at each end of the chromosome, such that the reported breakpoints could be at either end of the genome. Additional low frequency duplications were detected in deep sequencing reads of parental and adapted virus genomes (Table S5).

Figure 6. Evolution of transient copy number amplification

(A) β-gal activity from BHK cells infected with parent virus (ΔE3L) and replicates A-C at passage 10. Data are represented as mean +/− SEM. (B) Southern blot of the K3L locus from plaque-purified vaccinia viruses obtained from serial plaque purification of passage 10 replicate C virus in BHK cells. The presence or absence of the H47R substitution in K3L is indicated for each strain. (C), Southern blot of multi-copy plaque-purified virus (clone 5* from panel B) after four additional passages (+4p) in BHK or HeLa cells. Viruses passaged in HeLa cells showed re-expanded K3L copy number relative to those passaged in BHK cells.

Figure 7. Genomic accordion model of poxvirus evolution

The model depicts Red Queen conflicts proceeding through ‘gene-accordion’ intermediates. Gene copy number amplification provides an immediate fitness advantage and additional gene copies for sampling potentially beneficial mutations. After acquisition of a beneficial mutation, virus genomes are selected that retain advantageous point mutations without the gene expansion, leading to contractions down to a single gene copy.