Fig 1.
High-throughput single-cell analysis of the VACV infection cycle.
A, B, HeLa cells, cultured in 12-well plates at low density, were infected with Vaccinia virus (VACV) expressing two fluorescence reporters regulated by Early (pE/L) and PR (pA3L) promoters, enabling monitoring of early infection events, replication, and lysis. C, Widefield fluorescence microscopy images were segmented by a machine learning classifier using phase contrast channel data, generating masks which were applied to fluorescence channel images. D, Quantification of normalized fluorescence intensity (NFI) per cell and temporal tracking of single-cell infection events (n = 50 cells randomly sampled). Scale bars: 200 μm (A, C); 10 μm (B).
Fig 2.
Between-cell variability in VACV infection dynamics and outcomes.
A, B, HeLa cells were infected with VACVRep at an MOI of 1 and the normalized fluorescence intensity (NFI) of individual cells over time was quantified (n = 91). Means calculated from all infections analyzed indicated a typical progression of VACV infection. C, D, The underlying temporal dynamics at the single cell level revealed substantial between-cell variability (n = 50 randomly selected cells). Sigmoidal or double-sigmoidal models were fitted to single-cell fluorescence data, categorizing infections as non-lytic or lytic (based on Early NFI), and productive or non-productive (based on PR NFI). Percentages beneath each model represent the proportion of cells in that category. E, F, Key parameters describing single-cell infection dynamics were extracted from the fitted models. G, Comparison of infection start time between non-productive (NP) and productive (P) infections. H, Comparison of the timing of lysis between P, NP, and All infections. I, Comparison of the frequency of lysis between P, NP, and All infections. Points represent the mean of a field of view, while shapes represent replicate wells. Statistical comparisons were performed using unpaired Student’s T-tests. Comparisons with no annotation of significance shown were not significantly different. Error bars represent SEM (A, B).
Fig 3.
Correlations of key parameters describing VACV infections.
Parameters extracted from models fitted to VACVRep infections at an MOI of 1 were evaluated via pairwise correlations (Pearson’s correlation coefficient, upper triangle) and density distributions ( = mean, diagonal) for total (n = 40), lytic (n = 27), and non-lytic (n = 13) cell populations. Density indicates the relative frequency of values within the dataset. Parameters demonstrating strong correlation likely share an underlying mechanism or reflect the same facet of the host-virus interface. Statistical comparisons were performed using unpaired Student’s T-tests.
Fig 4.
Modulation of VACV infection outcomes by varying MOI.
A, Frequencies of infection (U-to-E; uninfected to Early gene expression), progression to productive replication (E-to-L; Early gene expression to PR gene expression), and lysis (E, L-to-lysis) for cells exposed to VACVRep at MOIs of 1, 10, 50 and 100. Points indicate the average of a field of view, while point shapes indicate replicate wells. B, Distributions of temporal delay values between initiation of Early and PR reporter expression for individual infections at MOI 1, 10, 50 and 100. C, Parameters extracted from models fitted to single-cell fluorescence data (MOI 1; n = 88 cells, MOI 10; n = 232 cells, MOI 50; n = 205 cells, MOI 100; n = 200 cells) were analyzed via pairwise correlations (Pearson’s correlation coefficient, upper triangle) and density distributions ( = mean, diagonal). Density indicates the relative frequency of values within the dataset. D, Rate of Early NFI accumulation at MOI 1, 10, 50 and 100. E, F, G, Two groups of parameters separated by a stronger (Timing parameters: Start Early, Start PR, and Midpoint PR) or weaker (Production parameters: Slope PR, Period PR, Max PR) response to change in MOI were compared by Multidimensional Scaling. Statistical comparisons were performed using Student’s T-tests (B, C, D), or chi-squared tests (A). Comparisons with no annotation of significance shown were not significantly different.
Fig 5.
Impact of inhibition of programmed cell death on VACV infection dynamics.
HeLa cells were infected with VACVRep at an MOI of 5, with or without 50 μM Q-VD-Oph (QVD), a pan-caspase inhibitor, and with and without 40 μM Necrostatin-1, a RIPK1 inhibitor. A, Frequencies of infection (U-to-E), progression to productive replication (E-to-L), and lysis (E, L-to-lysis). Points indicate the average of a field of view, while point shapes indicate replicate wells. B, Distributions of temporal delay values between initiation of Early and PR reporter expression for individual infections in the presence and absence of QVD. C, Density distributions of single-cell infection parameters were compared between QVD (n = 81 cells) and all vehicle only infections (n = 76 cells), as well as QVD and vehicle non-lytic infections only (n = 66 cells). D, E, F, Plaque assay performed on HeLa cells treated with QVD (50 μM) versus a vehicle control. One representative well is shown (D). Statistical comparisons were performed using unpaired Student’s T-tests (B, C) or chi-squared tests (A, E, F). Comparisons without significance annotation shown were not significantly different.
Fig 6.
Viral versus host driven infection dynamics.
A, B, VACV and PV single cell infection parameters rescaled by maximum normalization at a range of MOIs. Data first shown in Fig 4C. C, In our model, we propose that viral factors primarily dictate the timing of the early and PR VACV infection phases. Once the PR phase is initiated, the dynamics of virus production and assembly are limited by the host cell and are insensitive to the initial stoichiometry of infection. D, Poisson distributions showing the distribution of the number of infectious units (PFU) predicted to initiate an infection at the indicated MOIs. E, Coefficients of variation of selected single cell infection dynamics parameters for VACV and PV. Data first shown in Fig 4C. ‘Start’ for VACV represents the parameter Start Early. PV data was sourced from [4].