Live-Cell Imaging of Vaccinia Virus Recombination

Abstract

Recombination between co-infecting poxviruses provides an important mechanism for generating the genetic diversity that underpins evolution. However, poxviruses replicate in membrane-bound cytoplasmic structures known as factories or virosomes. These are enclosed structures that could impede DNA mixing between co-infecting viruses, and mixing would seem to be essential for this process. We hypothesize that virosome fusion events would be a prerequisite for recombination between co-infecting poxviruses, and this requirement could delay or limit viral recombination. We have engineered vaccinia virus (VACV) to express overlapping portions of mCherry fluorescent protein fused to a cro DNA-binding element. In cells also expressing an EGFP-cro fusion protein, this permits live tracking of virus DNA and genetic recombination using confocal microscopy. Our studies show that different types of recombination events exhibit different timing patterns, depending upon the relative locations of the recombining elements. Recombination between partly duplicated sequences is detected soon after post-replicative genes are expressed, as long as the reporter gene sequences are located in cis within an infecting genome. The same kinetics are also observed when the recombining elements are divided between VACV and transfected DNA. In contrast, recombination is delayed when the recombining sequences are located on different co-infecting viruses, and mature recombinants aren't detected until well after late gene expression is well established. The delay supports the hypothesis that factories impede inter-viral recombination, but even after factories merge there remain further constraints limiting virus DNA mixing and recombinant gene assembly. This delay could be related to the continued presence of ER-derived membranes within the fused virosomes, membranes that may once have wrapped individual factories.

Fig 1. Characterization of the recombinant VACV constructed for this study.

(A) Four recombinant viruses were constructed encoding combinations of the mCherry and Cro genes with or without a synthetic early-late pox promoter (pE/L), all inserted into the TK locus of WR VACV. For convenience, these are shown in the conventional orientation, but the inserts [with the exception of the pE/L-mCherry(dup) virus] are actually inverted relative to the virus genome. Recombinants were selected using mycophenolic acid resistance and/or mCherry fluorescence. (B) Subcellular localization of fluorescent labels relative to viral factories. EGFPcro BSC-40 cells were infected at an MOI = 5 with the indicated viruses. At the indicated times, the cells were fixed and stained for total viral and cellular DNA using DAPI. Images were collected using a spinning disc confocal microscope at 60× magnification. The scale bar = 15 μm.

Fig 2. Tracking the appearance of virus-encoded mCherry proteins.

(A) EGFPcro BSC-40 cells were infected at a MOI = 5 with VACV-pE/L-mCherry-cro, and the red and green fluorescence then tracked over time, collecting images 5 minutes apart, across 10 different fields (only a single representative field is shown). These are stills, the complete time-lapse movie is found in S3 Video. (B) EGFPcro BSC-40 cells were co-infected at a total MOI = 5 with VACV-pE/L-mCherry(t) and VACV-mCherry-cro viruses, and tracked using live cell microscopy to detect the appearance of recombinant mCherry-cro protein. The panels show different stills taken from S4 Video. Note the delay in the appearance of a mCherry-cro signal compared to panel (A). The scale bar = 25 μm.

Fig 3. Timing of expression of VACV post-replicative and late genes.

(A) EGFPcro BSC-40 cells were infected at a MOI = 5 with a virus encoding mCherry-tagged to the I1 protein (VACV-I1L-mCherry) and then tracked via live cell microscopy. These are stills taken from S5 Video. (B) mCherry-cro BSC-40 cells were infected and imaged as in (A) except using a YFP-tagged A5 virus (VACV-A5-YFP). These are stills taken from S6 Video. The I1L and A5L genes are post replicative and late genes, respectively. The scale bar = 25 μm.

Fig 4. Timing the appearance of early (I3) and late (A34) genes by western blotting.

EGFPcro BSC-40 cells were cultured in 60 mm dishes, infected with the indicated viruses [(A) VACV-pE/L-mCherry-cro; (B) VACV-pE/L-mCherry(t) + VACV-mCherry-cro], and then different dishes were harvested at the indicated times. Protein extracts were then prepared and western blotted for the indicated proteins. In parallel, the same viruses were used to infect EGFPcro BSC-40 cells on Fluorodish slides, transferred to a confocal microscope, and imaged over time. T_i is the time of initiating infection (in both arms of the experiment) and t_f is the time from factory formation, determined microscopically. Although I3 and A34 appear with essentially identical early and late kinetics, respectively, in both infections, the mCherry signal is greatly delayed in cells infected with VACV-pE/L-mCherry(t) + VACV-mCherry-cro viruses and appears only after A34 expression is first detected.

Fig 5. Timing of intraviral and virus-by-plasmid recombination events.

(A) EGFPcro BSC-40 cells were infected with VACV-pE/L-mCherry(dup) at MOI = 0.5 to favor infections with a single particle. Two different cells are tracked here from the time of factory development: one infected by an actively recombining virus (t_f ^a = 0:00, panel d), and another presumed to be infected with a “pre-recombined” virus (t_f ^b = 0:00, panel g). The appearance of mCherry expression in the cell infected with the pre-recombined virus (t_f ^b = 0:40, panel k) mimics that seen in cells infected with the pE/L-mCherry-cro virus (Fig 2A, t_f = 0:35), while the actively recombining virus produces a mCherry-cro signal late in infection (t_f ^a = 3:20, panel n). (B) EGFPcro BSC-40 cells were transfected with linearized pmCherry-cro plasmid DNA 4 hours prior to infecting with VACV-pE/L-mCherry(t) at MOI = 5. Images were collected immediately after initiating the infection and the appearance of new EGFPcro labeled DNA was used to track factory development (t_f = 0:00; T_i = 4:05) while mCherry fluorescence was used to detect plasmid-by-virus recombination. Images were collected every 5 minutes up to 10 h post-infection and assembled into time-lapse movies (see S7 Video and S8 Video). The scale bar = 25 μm.

Fig 6. Summary of different reporter protein expression kinetics.

The plot shows when a mCherry-cro signal is first detected relative to the time when a factory is first detected in that cell (t_f). Each imaging experiment was repeated 3 times, and 4/10 fields in each experiment were analyzed in detail, to produce the 12 data points per infection that are shown here. The data show that the appearance of a mCherry-cro signal is significantly (****, p <0.001) delayed in cells co-infected with pE/L-mCherry(t) and mCherry-cro viruses, compared to any other kind of infection. Note that pE/L-mCherry(dup) infections show two patterns of gene expression, some virus produce an mCherry signal shortly after factories are detected, and others produce a signal delayed by ~3 h. The VACV-pE/L-mCherry(t) + pmCherry-cro experiment refers to cells where a promoterless mCherry-cro plasmid was transfected into cells infected with VACV-pE/L-mCherry(t). Note that only a single experiment was used to produce the 12 data points collected for the I1L-mCherry and A5-YFP infections, these serve as timing reference points for post-replicative and late VACV genes, respectively.

Fig 7. Quantifying the recombinants formed in co-infected cells.

(A) The panel shows the two recombinant viruses that could be formed in cells co-infected with pE/L-mCherry(t) and mCherry-cro viruses. Also shown are the diagnostic restriction fragments that would be produced following XhoI digestion and the probes used for Southern blots (see Fig 8B). (B) Western blot analysis of proteins extracted from VACV-infected cells. BSC-40 cells were infected with the indicated viruses at a MOI = 5 (unless otherwise noted), harvested 24 h post-infection, and western blotted to detect recombinant mCherry-cro protein. A VACV gene product (I3) served as a marker of infection, and β-actin as a loading control. Note that both mCherry and mCherry-cro proteins are detected in cells infected with the control pE/L-mCherry-cro virus, suggesting that either could probably serve as a marker of recombination.

Fig 8. Southern blot analysis of recombinants.

BSC-40 cells were infected (or co-infected) with the different indicated viruses, and Southern blotted to detect recombinants. (A) The scheme used to collect samples for Southern blotting. Some DNA was extracted directly from virus-infected cells 24 h after infection. Alternatively, the viruses were plated, red fluorescent plaques subjected to two rounds of plaque purification, and the virus expanded to produce sufficient DNA for Southern blots. (B) Southern blot analysis of virus DNAs. DNA was extracted from virus-infected BSC-40 cells, as in panel (A), digested with XhoI endonuclease, and blotted using biotin-labeled pE/L and cro probes (Fig 7A). A novel 0.8 kbp DNA fragment is diagnostic for recombinants (and is also found in the pE/L-mCherry-cro control virus), but this 0.8 kbp fragment is only detected after the recombinant viruses are subjected to several additional rounds of plaque purification.

Fig 9. Recombination between pE/L-mCherry(t) and pE/L-mCherry-lacZ viruses.

(A) The figure shows the two parent viruses, the predicted recombinants, and the HindIII fragments that should be detected by the pE/L oligonucleotide probe (blue bar). (B) Southern blot analysis of cells co-infected with the pE/L-mCherry(t) and pE/L-mCherry-lacZ viruses. BSC-40 cells were infected with each of the parental viruses at MOI = 5, or co-infected with the two viruses at a combined MOI = 5, and the DNA was extracted 24 h post-infection. The samples were then Southern blotted using a biotin-labeled probe. Although faint, two bands at 2.2 and 0.9 kbp are seen that indicate the presence of recombinant genomes. Collectively they comprise about 2% of the DNA.

Fig 10. Maintenance of factory boundaries at an early stage of co-infection.

BSC-40 cells were co-infected with pE/L-mCherry-cro and pE/L-EGFP-cro viruses (schematics, top) for 4 h and then fixed and stained with DAPI to also detect virus and cell DNA. The images shown here are taken from a single Z-stack showing closely associated factories, one labeled with EGFP-cro and the other mCherry-cro. The images were collected with a spinning disk microscope at 60× magnification. See S9 Video for an alternative view of the image. The scale bars are 15 μm (top), 5 μm (middle), and 1 μm (bottom panel).

Fig 11. Large late viral factories enclose internal ER membranes.

(A) BSC-40 cells were infected with VACV strain WR for 4 h (middle panel) or 8 h (third panel) and then fixed and stained to detect DNA (DAPI), the ER membrane marker calreticulin, and the viral I3 protein. The images in Panel A show a projection of the Z-stacks. At early time points (4 h post-infection) one sees no calreticulin staining within the small early virus factories, although it is widely distributed throughout the cytoplasm. At later times, however, calreticulin-positive ER membranes appear to traverse these large late viral factories. This feature is more readily seen in an enlargement of the factory area, shown in the bottom row. (B) The same region of the image was separated into the component Z-stacks, these serial sections show the ER membranes extending downwards, through the factory. See S10 Video for an alternative view of the image. These images were acquired using an Olympus IX-71 inverted microscope at 60× magnification and deconvolved using Softworx software (GE Healthcare). The scale bar = 15 μm and each Z-stack spans 200 nm.