Table 1.
Comparison of key parameters of IAV replication in adherent MDCK and A549 cells.
Fig 1.
Virus release dynamics in response to in silico manipulation of gene expression of host cell factors in A549 cells.
We assume that the efficiency of individual steps in the virus life cycle is directly dependent on host cell factors and their influence is changed upon knockdown or overexpression of the corresponding gene. We simulated manipulation of gene expression by perturbing the corresponding kinetic parameters of an IAV replication model for A549 cells, which is based on a model previously established by our group [12]. For a simulated infection at MOI 1, virus release of the parental A549 (blue solid line) and the engineered cell line (brown solid line) are shown for the most important virus replication steps. Colors indicate whether perturbation of the corresponding step improved virus yield at 24 h p.i. by at least two-fold (green), had only an impact on the starting time point of virus release (yellow) or no impact (red). Scheme of IAV replication adapted from [22].
Fig 2.
Intracellular replication dynamics in response to in silico modifications of host cell gene expression in a single infected cell.
Changes in levels of viral cRNA (column 1), vRNA (column 2), mRNA (column 3) and matrix protein 1 (M1, column 4) are shown for a simulated infection at MOI 1 for the parental A549 cell line (blue solid line) or upon targeting selected steps of virus replication, as indicated on the left-hand side, by either knockdown (brown solid line) or overexpression (green dashed line).
Table 2.
HA titers and ranking results of cell lines overexpressing single host cell genes infected with A/PR/8/34 (H1N1) at MOI 10−4.
Fig 3.
Nuclear import of viral genomes in different A549 cell lines.
Model fit (lines) to experimental data (circles ± standard deviation, n = 4) for the import of viral genomes (vRNPs) in cycloheximide-treated cell lines upon infection by A/PR/8/34 (H1N1) at MOI 50. Relative increase in fluorescence intensity (FI) of the nucleus was determined by imaging flow cytometry after co-staining of cells with DAPI and vRNP antibody. The nuclear import rate was estimated by fitting the simulated fraction of nuclear vRNPs to the averaged experimental data. To account for the background signal of the nucleus in images, an offset of approximately 50% at 0 h p.i. was applied with respect to the experimental data obtained for parental A549 cells (A-F, blue), the transduction control (A, brown) and engineered cell lines overexpressing one of the following host cell factors: CEACAM6 (B, brown), FANCG (C, brown), NXF1 (D, brown), PLD2 (E, brown), XAB2 (F, brown). The asterisks indicate differences in relative FI levels with respect to the parental A549 cell line, that were either noticeable, however, statistically not significant (* p ≤ 0.1), or significant (** p ≤ 0.05) determined by Kruskal-Wallis test.
Fig 4.
Intracellular dynamics of viral RNA synthesis in different A549 cell lines.
Model fit (lines) to experimental data (circles ± standard deviation, n = 4 or single circles for FANCG, n = 2) of viral mRNA (panel 1), cRNA (panel 2), vRNA (panel 3) of segment 5 (encoding NP) in cell lines infected with A/PR/8/34 (H1N1) at MOI 50. Viral RNA synthesis rates and M1 binding rate were estimated by fitting the simulated number of the three viral RNA species to averaged segment-specific RT-qPCR data. To account for the offset in vRNA measurements caused by free viral RNAs in the seed virus we also implemented such offsets in our simulations with respect to the measurements obtained for parental A549 cells (A-F, blue), the transduction control (A, brown) and engineered cell lines overexpressing one of the following host cell factors: CEACAM6 (B, brown), FANCG (C, brown), NXF1 (D, brown), PLD2 (E, brown), XAB2 (F, brown). The asterisks indicate differences in RNA levels with respect to the parental A549 cell line, that were either noticeable, however, statistically not significant (* p ≤ 0.1), or significant (** p ≤ 0.05) determined by Kruskal-Wallis test.
Fig 5.
Virus particle release of different A549 cell lines.
Model fit (lines) to cell-specific numbers of released virions estimated from HA titer and maximum viable cell count (circles ± standard deviation, n ≥ 4) obtained from A/PR/8/34 (H1N1) infections at MOI 1. Simulated number of released virions was fitted to averaged cell-specific yield obtained for parental A549 cells (A-F, blue), the transduction control (A, brown) and engineered cell lines overexpressing one of the following host cell factors: CEACAM6 (B, brown), FANCG (C, brown), NXF1 (D, brown), PLD2 (E, brown), XAB2 (F, brown). The asterisks indicate differences in cell-specific yield with respect to the parental A549 cell line, that were either noticeable, however, statistically not significant (* p ≤ 0.1), or significant (** p ≤ 0.05) determined by Kruskal-Wallis test.
Fig 6.
Comparison of parameter values for viral kinetics obtained for A/PR/8/34 (H1N1) infections in different A459 cell lines.
After fitting 3000 resamplings of the available experimental data, all parameter values were normalized to the median of each kinetic parameter obtained for parental A549 cells. Bars represent the normalized medians and error bars indicate the first and third normalized quartile of each parameter per cell line (detailed boxplots in S2 Fig). The asterisk indicates differences in parameter values with respect to the parental A549 cell line, that were noticeable, however, statistically not significant (p ≤ 0.1) as calculated by a one-sided Z test (Gauss test).
Fig 7.
Evaluation of the time point of first virus release and the fold change in virus yield for model predictions of cell lines overexpressing multiple and single host cell genes.
Multiple gene overexpression cell lines (MGOs) were generated in silico by random assembly of kinetic parameter sets based on experimental single gene overexpression cell lines (SGOs), where letters in the upper right corner indicate which gene combinations were simulated (genes names are abbreviated as their first letter). For the resulting MGOs (pink dots) every 10th of approximately 1 x 104 - 2 x 104 model predictions are shown and compared to simulations with parameter sets experimentally determined for SGOs and parental A549 cells (dots, colors according to legend) at 48 h p.i. for a simulated infection at MOI 1, cell-specific virus yields were normalized to the one obtained for parental A549 cells. Open circles represent single cell predictions using the indicated optimal parameter according to the analysis shown in Fig 1. Dashed lines in histograms indicate the arithmetic mean of the corresponding simulation readout.
Table 3.
HA titers and ranking results of cell lines overexpressing multiple host cell genes infected with A/PR/8/34 (H1N1) at MOI 10−4.