Fig 1.
Map illustrating the structure of VACV genomes and the elements discussed in the text.
Panel A shows the 125 and 54 bp repeats that lie in-between the hairpin ends of the genome and the first and last open reading frames. “Native” viruses also encode variable numbers of 70 bp repeats that were omitted from the synthetic viruses studied here. The concatemer resolution site (CRS) is a highly conserved element that is required to convert concatemeric replication products into monomer genomes during packaging. All of these elements are embedded within inverted terminal repeats (ITR). Panel B shows the strategy used to assemble the virus. Different forms of hairpin ends were ligated to the left and right ITRs.
Table 1.
Oligonucleotides used for hairpin assembly.
Fig 2.
Structure, not sequence, determines the functionality of poxvirus hairpins.
Panel A shows the predicted structures of an Orthopoxvirus (VACV) and a Leporipoxvirus (Shope fibroma virus—SFV) hairpin end. The CRS is shown in black font and the sequence, which strongly resembles a poxvirus late promoter, is highly conserved between the two viruses. (Table 1 shows the sequences used to construct the two viruses.) Panel B shows the growth of viruses where the VACV genome was assembled encoding either VACV or SFV hairpins. Each virus titer was determined triplicate on BSC40 cells and error bars are shown where they extend beyond the data points.
Fig 3.
Mutant hairpin elements used in this study.
Panel A shows the different hairpin oligonucleotides that were used to try and reactivate VACV. The parent (S) hairpin comes from VACV strain WR. It took two attempts to rescue the SΔ1Δ3–6 tagged virus and three of the hairpins could not be rescued into infectious viruses despite three attempts at reactivation with each. Panel B shows an agarose gel that illustrates the different sensitivities of these oligonucleotides to mung bean exo/endonuclease. Mung bean nuclease is expected to cut at single-stranded sites in the molecules including at the hairpin ends.
Fig 4.
Growth and replication of VACV strains bearing mutant telomeres.
Panel A shows multi-step growth curves for each of the indicated VACV strains. Deleting a few of the mismatched loops had little effect on virus growth, but a mutant virus encoding only one of the original loops (SΔ1Δ3–6) showed a significant growth defect. Panel B shows a one-step growth curve comparing DNA replication in cells infected with viruses bearing the wild-type (S) or mutant (SΔ1Δ3–6) telomeres. The number of genome copies was determined using whole cell extracts, qPCR, and E9L gene probes. No obvious defect in DNA synthesis was detected in cells infected with the mutant virus.
Fig 5.
Southern blot showing telomere resolution in VACV-infected cells.
The assay detects the conversion of a 4.5 kb concatemeric (line form) replication intermediate into a 2.25 kb hairpin-ended DNA. DNA was extracted from infected cells at the indicated times, digested with Alw44I, fractionated by agarose gel electrophoresis (panel A), and Southern blotted using a biotin-labelled VACV telomere probe and chemiluminescent detection kit (panel B). A luminance imager was used to determine the signal intensities which were then used to calculate the amounts of telomeric virus DNA (panel C) and the ratio of lineform-to-hairpin DNA (panel D). Panel D shows two non-linear curves fitted using unconstrained one-phase decay models (R2≥0.95). There was no significant difference between the two calculated half-lives (panel D, inset) and no apparent defect in concatemer resolution in cells infected with the SΔ1Δ3–6 virus.
Fig 6.
Measurement of particle-to-PFU ratios.
Panel A. Flow virometry. Stocks of the sVAC-wt and sVAC-Δ1Δ3–6 viruses were grown and purified in parallel using centrifugation through a sucrose cushion. Ten-fold serial dilutions of each virus were prepared, fixed, and analyzed using a Cytoflex cytometer. Some aliquots were separately stained with SYBR dye. VACV particles in the >180 nm size range were counted using either SYBR dye fluorescence plus light scattering (dashed lines), or using light scattering alone (solid lines). For each data set a non-linear best fit to a sigmoidal curve was calculated using Graph Pad Prism. The figure illustrates the virus titer that corresponds to a particle count of 108 particles/mL (dotted line). Panel B. DNA content. Ten-fold serial dilutions of sVACYFP-A5-wt and sVACYFP-A5-Δ1Δ3–6 viruses were prepared and a 5 μL aliquot added directly to each qPCR assay mix. The number of copies of E9L was determined from known quantities of template amplified in parallel. For each data set a best fit line to the log-transformed data was calculated assuming both slopes equaled one. The figure illustrates the number of PFU associated with 103 gene copies (dotted line).
Table 2.
Vaccinia viruses used in this study.
Fig 7.
Transmission electron micrographs.
BSC-40 cells were infected with the indicated viruses at MOI = 3 for 24 hr and then fixed, processed, and imaged. Panel A. Low-magnification view of a cell infected with the sVACA5-YFP-Δ1Δ3–6 virus. White boxes show regions enlarged in panels B and C. Most of the particles appear to be immature virus (IV) although assembly intermediates are seen adjacent to regions of viroplasm (C). Panel D shows another cell again filled with IV, the area boxed in white is shown enlarged in panel E. Panel F shows a rare example of a particle resembling a mature virus MV (asterisk). Panel G. Low magnification view of a cell infected with the sVACA5-YFP-wt control virus. The area boxed in white is shown enlarged in panel H, where a mix of IV and MV particles can be seen near the viroplasm. Panel I shows another site of virus assembly, the IV particles seen in this and other fields closely resemble those seen in sVACA5-YFP-Δ1Δ3–6 infected cells. Panel J shows another cell filled with predominantly MV forms, this is more clearly seen in the enlargement in panel K. Panel L shows a field of MV imaged at a magnification like that in panel F.
Fig 8.
Transmission electron micrographs of virus particles.
Concentrated samples of the sVACA5-YFP-Δ1Δ3–6 and the sVACA5-YFP-wt control virus were fixed and embedded in agarose and then processed and imaged as described in Fig 7. Total particle counts in each field (~190 sVACA5-YFP-Δ1Δ3–6 and ~420 sVACA5-YFP-wt) were estimated using Fiji. The sVACA5-YFP-Δ1Δ3–6 virus specimen contained small numbers of oval particles exhibiting an appearance characteristic MV (arrowed) and shown enlarged inset. The sVACA5-YFP-wt control also contains a few spherical particles resembling IV (arrowed and boxed), although given the light contrast it’s possible to confuse these with some orientations of MV.
Fig 9.
Reduced proteolytic processing of the A3 (p4b/4b) protein.
The indicated quantities of each purified virus stock were denatured in loading buffer and size fractionated using SDS-PAGE. The separated proteins were western blotted to detect the p4b precursor and the cleaved 4b product. The amounts of virus loaded in each lane were calculated from the virus titers and because of the 10-20-fold excess of particles in the sVACYFP-A5-Δ1Δ3–6 virus stock (Fig 6), we see a corresponding excess of p4b/4b protein in lanes 8–10 compared to the controls in lanes 2–4 and 5–7. We compared the band intensities in lanes bearing comparable amounts of p4b/4b protein (lanes 2, 6, and 10). This analysis showed that >90% of the p4b/4b protein is cleaved to the 4b form in preparations of the two viruses bearing wildtype hairpins, but only ~30% of the protein is processed to the 4b form in the sVACYFP-A5-Δ1Δ3–6 virus stock.
Fig 10.
VACV I1 protein binding to selected hairpin oligonucleotides.
Each binding reaction contained ~100 ng of recombinant I1 protein, 100 ng of fluorescently-labelled hairpin oligonucleotide (as indicated), 700 ng of a non-specific poly·d(IC) competitor, and 0–300 ng of unlabeled S hairpin. Once assembled, the DNA-protein complexes were electrophoresed to separate the unbound hairpins (i) from more slowly migrating DNA-protein complexes (ii and iii), and then scanned to image the labelled complexes. I1 binds to all of the hairpins we tested under these conditions, but the S hairpin more readily disrupts the I1+SΔ1Δ3–6 and I1+SΔ1–6 complexes, than the I1+S complex itself (upper panel). The distribution of the fluorescent labels in each of the these and additional EMSA experiments (including I1 interactions with the SΔ1–3 and SΔ3–6 hairpins) is also plotted to illustrate the binding curves (lower panel, average ± SD, N = 3 per data point).
Fig 11.
Disease in VACV-infected SCID-NCr mice.
Mice were challenged by tail scarification with the indicated doses of a clone derived from a sample of Acambis 2000 virus (cA2K, panels A and B) or the sVACNM-SΔ1Δ3–6 (“no marker”) virus (panel C). A PBS control experiment was also performed in parallel with the two infection trials, and has been plotted twice for reference. Virus dose had no significant effect on the differences in survival in the cA2K-treated mice with median survival ranging from 30–37 days (panel B). All of the mice infected with sVACNM-SΔ1Δ3–6 virus survived over 70 days of study (panel C). The J2R locus was restored in the no marker virus to facilitate comparison with the J2R+ cA2K strain.
Fig 12.
Effect of hairpin mutations on VACV virulence in immune-deficient mice.
SCID-NCr mice (5 per group) were challenged with 105 PFU of sVACNM-wt or sVACNM-Δ1Δ3–6 viruses by tail scarification. A PBS control study was also conducted in parallel. The median survival in mice infected with the sVACNM-wt virus was 29 days whereas no animals were euthanized in either the PBS or sVACNM-Δ1Δ3–6 groups over 70 days (panel B). The tail lesions were also photographed once per week during the experiment. Panel C shows the persistence of the lesions in representative images.
Fig 13.
Sequence of a virus telomere recovered from a sVACNM-Δ1Δ3–6 infected tail lesion.
Panel A. The virus titer was expanded without a plaque purification step, purified, and sequenced. Reads derived from the hairpin region were retrieved as described previously [19] and are shown aligned with the SΔ1Δ3–6 hairpin drawn in line form. Panel B. M-fold models showing the predicted structure of the SΔ1Δ3–6 virus hairpin and the hairpin encoded by the recovered virus. The new virus seems to have duplicated a 16 nt sequence (red text) present in the original SΔ1Δ3–6 virus hairpin (blue text), and inserted it into where the original hairpin loop was found. This alteration is predicted to produce an additional mismatch and decrease the stability of the hairpin.
Fig 14.
Vaccine properties of the sVACNM-Δ1Δ3–6 virus.
Immune competent Balb/c mice (5 per group) were infected or mock-infected with the indicated viruses by tail scarification on day zero, and then challenged with a lethal dose of intranasal VACV strain WR, 28 days later. Both synthetically-derived viruses provided an equal degree of protection again the lethal challenge.
Fig 15.
Redistribution of F- and S- poxvirus hairpin elements.
Structure (i) shows a simplified view of a VACV genome bounded by two identical hairpins. These bear mismatched base pairs (blue lettering, mismatches = “o”) in an inverted repeat configuration and represents a simplified illustration of one of the starting structures assembled in this study. DNA replication is expected to generate a linear concatemer [3] composed of only Watson-Crick base pairs (ii), which can subsequently refold into a pair of Holliday structures flanking the central genome (iii). The process of branch migration reforms the mismatched base pairs and creates the virus hairpins. In VACV-infected cells these structures would be cleaved by the A22 Holliday junction resolvase [5], but depending on how these pairwise symmetrical cuts (e.g., cuts 2+2) are oriented within and between the two Holliday junctions, the reaction should produce four different arrangements of the F- and S- ends.